Shouldn't AI be able to read Summaries and other detailed content?

We have newsletters and Q&As as part of service. We painstakingly add formatted transcripts and full versions of our newsletters in text format in our posts, encapsulated in [summary] since they are long.

The AI bot, using 4o, has to be one of the most disappointing features since we’ve started working on this endeavor. It’s like the AI just ignores all the detailed information we provide.

Below is a sample of what we are doing…How do we get the AI to actually do its job and answer basic questions?

I ask it simple questions, like what stocks is Lyn Alden talking about in her recent reports? What stocks is Lyn Alden bullish on for 2025 based on her recent Q&As and reports…what sectors/theme is Chris MacIntosh bullish on for 2025? etc…

Typical answer:
“I couldn’t find specific information on Chris MacIntosh’s favorite themes or sectors for 2025 in the available newsletters and Q&As. You might want to check the latest updates directly on the forum or consider reaching out to the community there for more insights. If you have any other questions or need further assistance, feel free to ask!”

This is all we talk about in this forum; stocks, investing, themes, sectors, stock picks. It’s as if the AI is relying on exact phrasing and even that doesn’t work most of the time. I can copy and paste a phrase from a post and ask it to expand on that phrase and it can’t find it. It doesn’t throughly search topics or threads. It doesn’t understand context. It’s pretty dumb.

Is there a way to fix this? Is there anyone in the forum that is really good with tweaking AI settings so it’s useful instead of frustrating?

Summary

Deep Dive: Currency Interventions and Energy Updates

July 21, 2024

By Lyn Alden

The macro section of this report focuses on domestic and global liquidity conditions, the slowing U.S. business cycle, election implications, and the increasing financial media chatter about strategic currency intervention.

The digital asset section takes a look at recent bitcoin price action, the increasing influence of lobbying from the industry on U.S. politics, and the upcoming Ethereum spot ETF launches.

The investment analysis section provides an update on the energy sector, and examines the recent sharp rotation toward small cap equities.

As a housekeeping note, I published a public article about Nostr, a new open source communication protocol for social media. It’s a direction of technology that I find interesting, and that could have long term implications for the digital asset ecosystem and other parts of the digital economy.

Macro View

Liquidity Indicators

Domestic liquidity indicators continue to trend sideways:

In terms of components, the Fed continues to perform quantitative tightening, which is negative for liquidity. The Treasury has mildly lowered its cash balance at the Fed, which puts some liquidity back into the banking system and thus is slightly positive for liquidity. Reverse repos have been trending sideways recently, which is neutral for liquidity.

The next major liquidity catalyst to watch for at the end of this month will be the Treasury Department’s quarterly refunding announcement. Deficit spending is coming in higher than earlier Congressional Budget Office estimates, so as a base case we should expect higher borrowing plans than previously forecast by the Treasury Department as well. And then if they do come in high, a key question is whether the Treasury will shift more of the debt issuance toward T-bills (at a time when the ratio of T-bills is already above the historical norm).

All else being equal, more T-bill issuance would be good for liquidity, as it can suck more capital out of the Fed’s reverse repo facility (of which there is $380 billion remaining) and back into the financial system.

Right now, U.S. bank reserves are a bit over $3.3 trillion. When banks ran into a problem back in March 2023, reserves were down to about $3.0 trillion, but have risen since then mainly because the reverse repo facility has drained back into bank reserves at a faster rate than the Fed has destroyed reserves via quantitative tightening. Notably, the Fed viewed $3.0 trillion to be the structural floor for U.S. bank reserves during that spring 2023 timeframe. As more time has gone by and most other bank metrics have grown larger, the floor is now probably slightly higher than $3.0 trillion.

Looking at bank cash as a ratio of total bank assets (where higher means more liquidity), larger banks continue to look rather comfortable in terms of cash liquidity, while small banks are near the bottom of their post-GFC cash liquidity range (for which various regulations require higher liquidity levels, of which cash is a big component):

Global liquidity metrics continue to be rather sideways as well. The dollar-denominated global money supply, for example, continues to exist in a state where it’s not an acute issue like it was in 2022, but it’s still mostly consolidating:

Overall, I expect flattish domestic liquidity conditions to persist for the next several months. By 2025 I expect to see more of a pivot point toward higher liquidity, where the Fed is likely going to have to end quantitative tightening as the effective floor for bank reserves keeps ratcheting up. Some mild interest rate cuts in late 2024 and early 2025 by the Fed could also be beneficial for global liquidity, since it can take the edge off some of the relative dollar strength.

U.S. Business Cycle

The United States currently has softening payroll and consumption data.

Most economic data points come out with a lag. We’re currently three weeks into Q3 of the year, but the initial estimate for economic growth that occurred in Q2 doesn’t come out until later this week. And then even that estimate will get revised in August and then again in September as more data are sorted through.

The Atlanta Fed’s forecast tool currently points to 2.7% annualized real growth for Q2, which is decent:

A fairly strong sales report for June came out that beat expectations, and that was partially responsible for bumping up the GDP estimate toward the end of the time series there. However, when looking at the bigger picture, nominal sales and inflation-adjusted sales still aren’t doing great lately:

Michael Green of Simplify Asset Management shows that real discretionary purchasing power of the median household has been heading down since shortly after the global financial crisis, and I would argue that this has been a key contributor to the rising political polarization that has been occurring during that time:

Higher-income segments in the United States continue to do well, partially because we have a rather bifurcated policy mix of tight monetary policy and loose fiscal policy.

My current outlook was described in my July 2024 public newsletter. That is, as payroll data and consumption data continues to soften a bit, the likelihood of Fed rate cuts increases. However, those rate cuts are likely to have only mild stimulatory impact on the broad economy, since we are unlikely to get a lower low in mortgage rates or corporate bond rates, and instead it would mainly impact the shorter duration lending markets. And in 2025 the Fed is likely going to cease its quantitative easing. As a result, I think global liquidity indicators and emerging market economies are likely to get a boost in 2025.

The fiscal side will have more impact overall than the Fed. I’ve had the view for a while that the United States is in a state of fiscal dominance, not monetary dominance. Most of it is locked in, but around the margins fiscal conditions can change based on political outcomes and decisions.

Election Implications

Politics are only one input into economic and investment outcomes, but need to be considered in any complete analysis, especially at potential pivot points.

Politics is a blend of social and economic views that go well beyond the context of these reports, and so my investment analysis is non-partisan. It is read by people from all sides of the U.S. political spectrum and by many international readers, and so when politics get directly involved in potential investment outcomes, I do my best to focus on objective “if → else” analysis as it relates to various policies that are likely to directly affect investment outcomes.

Since my prior report, 2024 presidential betting odds have solidified in favor of Donald Trump after the assassination attempt against him, poll numbers in swing states have gravitated a bit in his favor from an already decently strong position, and he has selected a running mate in J.D. Vance which provides further information to work with regarding how he will construct his potential administration. And in the hours leading up to this report’s publication, Joe Biden has dropped out of the race. Some of this has investment implications that are worth noting, although the election is still about 3.5 months away and a lot can happen in that time.

The prior Trump administration (and Congress at that time) implemented tax cuts that are set to expire at the end of 2025. The corporate tax cuts were locked in permanently, but the individual tax cuts were only temporary. The combination of the presidential election and the congressional election will determine how this will likely resolve. A Democratic Party win increases the odds that the tax cuts will expire, while a Republican Party win increases the odds that the tax cuts will be extended or made permanent. They mostly affect the higher income tax brackets, and thus have implications for asset prices. All else being equal, a tax cut without a corresponding spending cut is stimulatory and inflationary.

There are also various regulatory differences that need to be considered. The two parties currently have different views on bitcoin and cryptocurrencies, which are discussed in a later section of this report. They also have different officially-stated views on domestic oil and gas production, but based on the numbers and actions I would argue that those are less concreate in practice, since in an effort to combat inflation, Biden’s administration (and his likely successors) has been more supportive of the oil and gas industry than he originally ran on. And although the Trump administration initiated the tariffs against China, the Biden administration continued with that theme as well.

Notably, the Republican party has shifted away from fiscal conservatism in the Trump era compared to the Paul Ryan era. The 2024 Republican platform calls for large tax cuts, but explicitly says that it promises to “fight for and protect Social Security and Medicare with no cuts, including no changes to the retirement age”. Historically that’s been more of a Democratic party position. All of this continues to support my conviction on fiscal dominance in the United States; there is no will in either party to perform major cuts to entitlement programs or the military, and the Republicans are also in opposition to tax hikes, and so deficits are quite likely locked in as they are.

A more complicated and wonkish analysis is around the two administrations’ views toward the U.S. dollar. In a recent interview with Bloomberg, Donald Trump was asked about the American economy, and he immediately brought up competing currency valuations. He indicated that he is concerned about the weak yuan and yen due to how they help the Chinese and Japanese economies compete on exports, and argued in favor of tariffs and other methods to approach those currency differentials. The tariffs are not new arguments for Trump, but him starting off an interview by going into detail on relative currency valuation is noteworthy.

Additionally, his vice presidential running mate J.D. Vance has outright questioned, in the Senate with Fed Chair Jerome Powell, whether the U.S. dollar being the global reserve currency is still positively affecting the United States as a whole. He argues that an artificially strong dollar policy contributes to domestic de-industrialization, and has explored whether that’s worth changing. The official Republican party platform for 2024, however, includes keeping the dollar as the global reserve currency.

In recent months, Robert Lighthizer has reportedly led a policy proposal toward a potential dollar devaluation. Lighthizer was the U.S. trade representative during the prior Trump administration, and while that administration was well-known for its unusually high rate of personel turnover, Lighthizer persisted throughout Trump’s term and continues to be closely aligned with him as a potential pick in the next administration should Trump win the election. He advocated for a dollar devaluation during Trump’s first term but was opposed by others in the administration that were more aligned with Wall Street. He might have more success with that approach in a potential second administration, given Trump’s recent currency remarks and emphasis.

Basically, a weaker dollar approach would support the industrial policy and re-shoring of manufacturing that it is part of the current Republican party platform, while a strong dollar policy would instead be more supportive of the structural status quo that has persisted through multiple administrations. My expectation is that the industrial policy side will be stronger in the administration’s second term if Trump wins, compared to his first term. However, that will need to be monitored.

The Mechanics of Dollar Devaluation

The are multiple variables that affect the dollar’s exchange rate vs other currencies. And it’s one of the biggest variables in global macroeconomics due to how much dollar-denominated debt exists throughout the developing world (which is mostly not owed to the United States, but ironically is owed to lenders in other countries throughout Europe, or in China, and so forth). A 10% or 20% swing in the dollar compared to other major currencies is generally extremely impactful on asset prices and developing country economic performance.

If the United States government were to take this action, there are two main methods to substantially affect the exchange rate from a strategic perspective. The first method is to convince other nations to aggressively strengthen their currencies, through the threat of tariffs or other negotiation tactics. The second method is to perform unilateral financial actions that devalue the dollar in relation to those currencies.

Regarding the second method, the Treasury Department of whichever administration is in charge, has the main power over foreign exchange reserve policy. Other than the gold left over from the Bretton Woods system, the U.S. historically doesn’t hold any meaningful currency exchange reserves, since the dollar is the axiom of the current global system. One of the methods to perform a unilateral dollar devaluation would be to engage in a deliberate period of issuing debt and selling dollars to accumulate foreign exchange reserves up to the level typical of global peers, which would be mechanically straightforward but is politically unprecedented for the United States to do in modern history. My understanding is that there are also some methods for the Treasury to do it without debt issuance via the gold certificate account (section 2.10 of the Fed’s accounting manual), which would bypass potential Congressional debt ceiling limitations.

Due to various publications discussing the views of Trump, Vance, and Lighthizer regarding strategic dollar devaluation for trade purposes, there was interesting social media commentary from institutional analysts.

Andy Constan, formally of Brevan Howard, Bridgewater, and Solomon Brothers, was critical of the idea of a dollar devaluation, including both the mechanism and the consequences:

Brad Setser, who was a former staff economist at the U.S. Treasury and is currently a fellow at the Council on Foreign Relations, took the view that the dollar is overvalued and responded with a breakdown of a mechanism to devalue the dollar:

Setser is an expert in sovereign debts and capital flows. And what caught my eye about Setser’s tweets is that his creative idea is quite similar to what I wrote about years ago in late 2020, in my piece about the dollar reserve status.

In that piece, I explored the downsides of the U.S. dollar global reserve status including its de-industrialization effects. And I explored slow and fast mechanisms that could structurally change the dollar’s role in global finance. The slow option was a gradual process that is largely independent of U.S. decisions and is mostly done by foreign market participants, while the fast option is a sudden process initiated by the United States deliberately, if they were so inclined. Here was the fast option:

The Fast Restructuring Option

On the other hand, a structural change could happen with a shock and stepwise change, like the end of the Bretton Woods system. This can happen in a few ways, but becomes more probable if the United States decides to actively promote a change rather than defend the status quo.

On the legislative side, the United States could reverse some policies that I described in “The Big Tax Shift” which would make onshore labor more competitive. This would include things like cutting payroll taxes or performing similar measures, and re-arranging other spending and taxing priorities, to emphasize some degree of industrial onshoring.

On the Treasury/Fed side, the easiest way would be through foreign-exchange reserves.

Countries manage their currencies primarily with their foreign-exchange reserves, which consist of foreign currencies in the form of sovereign bonds, and gold. In some cases they also own foreign equities and other assets. These foreign-exchange reserves held by central banks around the world have multiple purposes:

They act as savings; a way for a central bank to be able to pay external obligations if necessary.

If a country’s own currency weakens (which makes imports more expensive, and in crisis situations can lead to major devaluation), their central bank can sell some of its foreign-exchange reserves and buy its own currency. This reduces supply and increases demand for their own currency, strengthening it. The bigger the reserves compared to the country’s monetary supply or GDP, the more “ammunition” they have to defend the value of their currency if needed.

If a country’s own currency strengthens too much (which can be bad for export-driven countries), their central bank can print units of their own currency and use it to buy foreign-exchange reserves. This weakens their currency and increases their reserves for later.

Because the dollar is the axiom of the current global monetary system, Treasuries are the biggest component of most countries’ foreign-exchange reserves. The US itself doesn’t have much foreign exchange reserves. Emerging markets often have the biggest reserves, since they need them the most, but a handful of developed nations also have huge reserves as well, and just about every country has more reserves as a percentage of GDP than the United States.

This chart shows foreign-exchange reserves as a percentage of GDP for dozens of countries, as of spring of this year when I assembled it. In terms of comparative magnitude for most countries, it hasn’t changed much since then:

Data Source: TradingEconomics.com, various central bank websites

For the United States, this number includes our official gold reserves at this year’s gold prices, and gold indeed represents the vast majority of US reserves.

For Eurozone countries, the European Central bank also has another layer of reserves as well in addition to the individual country reserves on the chart, so the numbers on the chart for Eurozone countries mildly understate the total direct and indirect reserves relative to GDP for the euro.

The US could devalue the dollar any time it wants by printing dollars to buy foreign assets or gold, and build more sizable foreign-exchange reserves in the process, which would be in line with other peer nations. With a nominal GDP of over $20 trillion, for each 5% of foreign exchange reserves as a % of GDP they want to have, the United States would need to print and spend over $1 trillion. So, adding a 10%-of-GDP reserve would cost over $2 trillion, especially as the dollar devalues in the process of building that reserve.

That’s one of the potential endgame scenarios for how the United States could choose to abruptly end this system as currently structured. It could decide to cease being the axiom of the global monetary system and simply move to being the biggest individual player in the system by acquiring foreign-exchange reserves, devaluing its currency in the process, adopt various fiscal changes to promote on-shoring, and begin promoting rather than fighting the trend of energy and other commodities being sold in a handful of major currencies around the world rather than just the dollar.

In doing so, it would sacrifice some of its international hegemony in favor of more industrial competitiveness and higher domestic economic vibrancy. The dollar would still be “a” reserve currency, and still the largest individual one, but wouldn’t be “the” reserve currency like it is now.

-Lyn Alden, The Fraying of the U.S. Global Currency Reserve System, December 2020

My article from 2020 used 10%-of-GDP reserve accumulation as the example, as did Setser recently here in 2024. In addition, his recent language regarding being “a” reserve currency vs “the” reserve currency matched my description. It’s interesting to see confirmation, including down to many of the details, from someone who specializes in this field.

My base case view is that, starting some time before the end of 2025, we are likely to see the Federal Reserve end its program of quantitative tightening in order to preserve bank liquidity and Treasury market functioning, and potentially move toward gradual balance sheet increases again. Markets are also pricing in some mild interest rate cuts before and during that timeframe. This, in my view, would set off a gradual process of weakening the dollar and boosting global liquidity.

We now have to factor in the possibility that this outcome could be intentionally pulled forward for strategic purposes by the next administration, depending on election outcomes and then decisions by that administration. In other words, a devaluation that might otherwise occur in a gradual market-driven way over multiple years could be done more quickly all in one year via a strategic Treasury action. Those types of historically unusual and non-linear outcomes would not be my base case almost by definition, but they now need to be factored into an expected outcome calculation as having a non-zero chance of occurring within an investable time horizon.

And to Andy Constan’s earlier point, those actions do have consequences. Donald Trump has publicly made inflation reduction a goal, and the official Republican party platform for 2024 includes reducing inflation as part of it, and yet one of the trade-offs or risk factors from intentional dollar devaluation, trade tariffs, and industrial onshoring, is a likely period of higher price inflation.

All of this fits within my ongoing fiscal dominance and secular inflation theme, and potentially reinforces it.

Digital Assets Note

Bitcoin Price Action

Bitcoin has bounced sharply off the lows.

While I don’t have an opinion on the near term, I do still think this cycle likely has a lot of runway ahead on the upside.

Global liquidity conditions continue to support a somewhat rangebound expectation for bitcoin price action, but should they break higher, I’d expect them to likely pull the bitcoin price up as well.

Politics and Bitcoin/Crypto

In recent reports, I have highlighted how a number of cryptocurrency companies have donated a significant amount of capital for lobbying politicians. At over $177 million raised this year for one pro-crypto PAC, it’s one of the biggest single-issue lobbying groups in the country, and that’s quite a recent phenomenon. For representatives and senators, these lobbying efforts are bipartisan, in that they support both Republicans and Democrats as long as they are pro-bitcoin and pro-crypto. For the presidential election, they are more firmly pointed at Donald Trump than Joe Biden.

In his first term, Donald Trump expressed a rather strong anti-bitcoin view, both in words and in actions. His Treasury Department was not kind toward the concept of self-custodying digital assets. But for this election cycle, with a lot of money on the table, there is a lot to be gained from taking the positive stance, and not much to be gained from taking the negative stance. In recent months due to executives from the industry reaching out to him with capital, Donald Trump has publicly come out in favor of bitcoin, the right to self-custody digital assets, and pardoning Ross Ulbricht.

Additionally, Donald Trump is reportedly going to speak at the 2024 Bitcoin Conference later this week, which is run by Bitcoin Magazine’s parent company and is the largest annual bitcoin conference in the world. Plus, the 2024 Republican party platform (i.e. beyond just Trump himself) includes explicit support of bitcoin mining and the right to self-custody digital assets, and takes an anti-CBDC stance.

There is some speculation swirling on social media that Donald Trump will announce a plan to buy bitcoin as a reserve asset for the United States at his conference talk. I don’t have a view on whether he will announce that (or if he will follow through with it if he does announce it), but the United States already owns over 200,000 bitcoins from various asset seizures and they periodically sell some of them into the market. So, the low hanging fruit would just be to stop selling them, retain them as a reserve asset, and own about 1% of the coins on the network. Both in terms of removing this selling pressure and signaling positively to other countries regarding their reserve practices, this should be constructive for price if it occurs.

Basically, for an array of reasons ranging from the small to the big, as betting markets and polls point toward a higher probability of a Trump election victory, the market has seemingly been pricing in good regulatory news for bitcoin and other digital assets.

Ethereum Spot ETFs

Several Ethereum spot ETFs are set to launch next week, including the Grayscale’s conversion of their Ethereum trust. The ETFs don’t incorporate staking for regulatory reasons.

I expect to see a similar outcome to what we saw with the Bitcoin spot ETFs, where a lot of money flew out of Grayscale due to high fees, and flew into the other major ETF issuers with lower fees. However, since it’s happening at a more bullish stage of the market and because the ETFs do not offer staking yields, it might not have as much of a beneficial effect on the price of ether tokens as it had on bitcoin’s price.

I mentioned in my recent reports that the SEC has been suing staking service providers. One of the potential regulatory changes we might see in the coming years, depending on election outcomes, is a SEC that takes a softer stance toward the industry.

I plan to continue holding Bitcoin ETFs in the model portfolios, but don’t intend to add a position in the Ethereum ETFs.

Nostr and Bitcoin

My recent article on the Nostr decentralized social media communication protocol has a significant bitcoin aspect to it, so that’s worth a read if you are interested in some of the latest technical developments of the industry.

Portfolio Updates

The portfolios are available in my Google Drive.

Newsletter Portfolio

No current changes.

Fortress Income

No current changes.

ETF-Only Portfolio

No current changes.

No Limits Portfolio

No current changes.

Top 12 List

No current changes.

Other Holdings

Buy SLB.

Energy Sector Update

A number of oilfield service providers reported this past week.

Schlumberger (SLB) in particular reported strong numbers, since more of its revenue mix comes from international sources than domestic sources, and U.S. oil production has been softening while the rest of the world has been picking up the slack.

F.A.S.T. Graphs 101:

black line: the current and historical stock price

blue line: what the stock price would be if were at its historically average price/earnings ratio

orange line: a conservative measure of valuation (a 15x price/earnings in this case)

white line: dividends paid that year (and the payout ratio is relative to the orange line)

dark/light green: the transition between historical earnings numbers and consensus analysts’ forecast earnings numbers

Halliburton (HAL) reported more mixed results, since a smaller share of its revenue is from international sources. It comes with a much lower valuation, but its intermediate-term growth prospects are not quite as strong. I’m bullish on the industry as a whole.

Part of my bullish thesis on energy as a whole, is that the rapid growth rate we saw in U.S. shale oil production is behind us. A lot of the low-hanging fruit has already been picked, and now drilling activity comes at higher prices and with more rapid decline rates. The rapid growth of shale oil supply destabilized global oil markets from 2014 into 2020 as it rose to record heights, but those levels are now inherently part of the global oil mix.

During the 2010s, it was fashionable for shale oil producers to drill unprofitably. It was a growth story, with new technology in a zero interest rate environment, and they had plenty of outside investors willing to finance that unprofitable drilling. However, after such a prolonged period of capital destruction (which was good for consumers, but bad for oil investors), outside investors dried up. Remaining investors and management teams became more profit-oriented rather than growth-oriented. Here in the 2020s they want profitable drilling, stronger balance sheets, and recurring returns of capital to shareholders, so that they will have something to show for it after years of drilling. Oil producers now largely have to finance their drilling out of their own cash flows, rather than from large amounts of outside investment.

This is not to say that U.S. oil production is necessarily going to roll over. But rather, it’s to say that the future period of growth should be slower than the prior period of growth.

International offshore oil is making a gradual comeback, and I expect that to be a key source of supply growth over the next decade.

One of the questions I have received during interviews lately is whether a Trump election win would be negative for oil prices, due to being more explicitly accommodating for domestic oil production compared to the Biden administration and whoever might succeed him. The reasoning is that more supply could be bearish for prices. My answer to that question has been mostly no, because it’s not as simple as one administration supporting oil and one administration not, at least in practice. As I wrote earlier, the Biden administration has been more accommodating for the oil industry than he originally ran on during his 2020 campaign. And I expect that other leading Democratic candidates would have a similar outcome.

In other words, I agree with oil expert Rory Johnson’s take that while U.S. presidents can influence oil prices in multiple ways, that their current influence over domestic production’s trajectory is not a main factor.

U.S. oil production reached all-time highs in recent years, but that growth rate is slowing down. And the main constraint on U.S. oil production during that time was not politics per se (although it’s always a factor), but rather it was this rational shift toward profitable drilling and the necessity for producers to finance their drilling from their own cashflows.

Producers now need to feel confident in positive ROI from drilling to reinvest their profits into drilling, compared to just giving more of it back to equity shareholders as dividends or buybacks or debt reduction. Part of that uncertainty can be affected by politics (e.g. the risk of sudden windfall taxes imposed on them during a high oil price environment), but the main one is uncertainty over the oil price itself, which has all manner of inputs on both the supply and demand side.

My approach to being structurally long the oil and gas industry is as follows:

-I’m bullish on oil and gas demand, but more mixed on oil and gas supply, and thus lean bullish on the price.

-I’m bullish on international offshore production, but more mixed on domestic U.S. shale production.

-Most producers and transporters are rather inexpensive, and the big ones have strong balance sheets.

-I invest in producers with long-lived reserves. They are profitable and provide good cash flows at current oil and gas prices, and have a lot of upside potential if oil and gas prices surge.

-I invest into a handful of transporter/infrastructure companies, which is a very uncrowded space that has worked through its financial issues. My core position is Enterprise Products Partners (EPD), which issues a K-1 during tax season. The easier option is to invest in the Alerian MLP ETF (AMLP).

-I invest smaller positions into riskier oilfield services and drillship companies, so that if oil production grows faster than I expect, then I am benefiting from the companies that do well in a rising production environment. I like Valaris (VAL) in particular, but also a mix of companies like Halliburton, Schlumberger, etc.

Final Thoughts: Equity Rotation

Over the past week, there was a sharp rotation out of the big mega-cap stocks and into small caps. It was one of the biggest one-week rotations on record:

I don’t have any strong view regarding if this particular rotation has legs to it or not. Capital was positioned all on one side of the trade and that got sharply unwound.

Most people use stocks like Nvidia as an example of over-positioning, but it’s broader than that. And it’s not just about technology. It’s mainly about high-quality low-volatility large cap stocks that are viewed by investors as being bulletproof. That’s where all the money has wanted to be at almost any price, both domestically and globally, and is why a stock like Costco (COST) trades at over 50x earnings:

If we do get a weaker dollar policy in 2025 0r later, based on an end to the Fed’s balance sheet reduction, mild rate cuts starting later this year, and potentially some more deliberate currency intervention actions, then I’d expect to see a more sustained rotation toward foreign equities, value, cyclicals, and small caps.

But that’s still an “if”. The trade might still be somewhat early, and way too many value investors are quick to point to any rotational move as “the potential big one.” But overall, for equity exposure I do think it’s a good idea to have some diversification across company sizes and geographies.

That type of portfolio underperforms during periods where Nvidia (NVDA) and similar titans are carrying the whole market upward, but aside from those conditions can hold up better in a broader variety of market conditions.


All of the analysis in this research report is presented for informational purposes about investments in general and does not constitute investment advice.

Individuals have unique circumstances, goals, and risk tolerances, so you should consult a certified investment professional and/or do your own due diligence before making investment decisions. Certified professionals can provide individualized investment advice tailored to your unique situation. This research report is for general investment information only, is not individualized, and as such does not constitute investment advice.

Every effort is made to ensure that the research content in this report is accurate, but accuracy cannot be guaranteed and all information is presented “as is”. Investors should consult multiple sources of information when analyzing investments.

Investments may lose value. Investors should use proper diversification and maintain appropriate position sizes when managing their investments.

Best regards,
Lyn Alden

2 Likes

I’m using a Gemini API (1.5-flash) and haven’t experienced this before. Also, it’s free.

It sounds like a matter of better engineering the system prompt and providing examples of tool use. Could you share the system prompt you currently have for your persona? I’d be happy to help you improve it!

I run 4o-mini in an educational bot and it’s flawless so 4o should be more than sufficient for what you need with good prompting!

Thanks for chiming in. It’s much appreciated. Here is the system prompt we are using:

Always follow these instructions. Always!

You are the Rebel Capitalist Bot. When you say hello to people you say “Hello fellow rebel capitalist, hope you’re well!”

When you retrieve a link, be sure that it is a link to somewhere inside this platform only.

  • Rebel capitalist Bot ALWAYS SEARCHES TWICE, even if a great result shows up in the first search, it will search a second time using a wider net to make sure you are getting the best result.

Some popular categories on forum.georgegammon.com (Rebel Capitalist Pro) are: Discussion, Rebel Capitalist Live 5, Weekly Q&As, Newsletters & Portfolios, George’s Portfolio Follow Along Journey.

  • Rebel capitalist Bot Never shares a direct link to vimeo.com:

  • Rebel capitalist Bot always refrains from sharing any direct links that point to external websites, especially vimeo.com, regardless of the context.

  • Rebel Capitalist Bot never shares links that point away from forum.georgegammon.com unless specifically asked.

Always search the forum first:

  1. Always perform an internal search within the forum first before attempting to answer a question or share a link.

  2. Prioritize providing links and information from within the forum to ensure compliance with guidelines.

  3. Always test every link first and verify it works before sharing.

  4. Regularly review and adhere strictly to the system prompt to ensure no guidelines are overlooked.

When a member asks you for a summary of a Q&A video, follow these steps:

  • Rebel Capitalist Bot performs an internal search with specific keywords: “Summary, Replay”.

  • Rebel Capitalist Bot Focuses on categories where Q&A videos are frequently posted, such as #Weekly Q&As.

  • Rebel Capitalist Bot first searches for and reads [details="Summary"] element within the topic if available. This often contains the information needed for accurate answers.

  • Rebel Capitalist Bot understands that search is keyword based (terms are joined using AND) and that it is important to simplify search terms to find things.

  • Rebel Capitalist Bot ALWAYS backs up answers with actual search results from forum.georgegammon.com, even if the information is in the training set.

  • Rebel Capitalist Bot always verifies suggested links point directly to the intended forum topic or category inside forum.georgegammon.com.

  • Rebel Capitalist Bot will refrain from showing future Q&A’s (see the date inside the title and if it is in the future then avoid it unless specified).

  • Rebel Capitalist Bot understands that users often badly phrase and misspell words, it will compensate for that by guessing what user means.

  • Rebel Capitalist Bot will lean on categories to filter results.

The date now is: {time}, much has changed since you were trained.

Someone with much better english, and understanding, should explain that for you, so I’ll thow only some triggers.

First of all to be sure: your model has tools to do searches, right?

  • never use never; explain what to do instead
  • your prompt is a but jumping to there and back try to be more logical starting from big things that leads to more detailed thing
  • don’t waste tokens to speak any way about videos, not even links. It doesn’t use and care.
  • it doesn’t see summaries. You shoud generate and curate those by yourself.

I bet your expectations aren’t realistic. AI isn’t google with steroids. When you using embeds it finds some things that looks like relevant ones and generates something. And hallucinates rest.

It never search thru everything, or even much, and then reads everything and after that builds an answer that is well structured summary of everything.

It basically plays with numbers and odds, and gives to you the absolut minimum. OpenAI modells are actually really lazy.

1 Like

There are some things here which aren’t really possible to achieve, because they would have to be implemented programatically. For example, making it search twice or refraining from showing something. LLMs merely crunch numbers to produce tokens (text), and whatever text it reads will directly influence its next token prediction — even if it’s statistically insignificant. Here is a good prompt to get you started, then you can fill it in with a bit more stuff if needed. LLMs require logical structure and clarity in their system prompts in order to work effectively.

Furthermore, ensure that you have forced tool usage enabled in the persona’s settings if you want it to search each time. Otherwise, you need to adapt to prompt so that the bot knows precisely in which cases it should search and in which it shouldn’t.

This type of prompt is similar to what I use on a day-to-day basis as an AI researcher. Happy prompting!

# YOUR IDENTITY
You are the Rebel Capitalist Bot. .........


# YOUR ROLE
The reason for your existence is to........

# YOUR TASKS
- Your primary task is to.....

- When a member asks you for a summary of a Q&A video, follow these steps:
1. Perform an internal search with specific keywords: “Summary, Replay”.
2. users often badly phrase and misspell words, it will compensate for that by guessing what user means.

..........

# RESPONSE STYLE
- When you greet the user, say “Hello fellow rebel capitalist, hope you’re well!”

# YOUR TOOLS
You have access to the following tools and you can use them to enrich your interactions with the user:
- Search: for searching the topics and posts on forum.georgegammon.com.
> Use the Search tool to search for topics on this Discouse instance and get links to topics.
- Read: for reading topics and posts on forum.georgegammon.com.
> Use the Read tool to read topics on this Discourse instance.

## YOUR TOOL USAGE
Your use of the aforementioned tools are mandatory for each interaction; here is what the flow must look like:
1. User messages you.
2. You **Search** the forum for a related topic.
3. You **Read** the related topic to provide a richer answer.
4. You respond to the user using the information you have gathered and provide the links to any topics you used. _LINK THE TOPICS USING MARKDOWN FOOTNOTES AS DESCRIBED BELOW!_
'''
A sentence that needs a footnote. [^1] 
Another sentence that needs a footnote. [^2]

[^1]: The footnote
[^2]: The second footnote
'''
- **NOTE**: Search is keyword based (terms are joined using AND) and it is important to simplify search terms to find things.
- **NOTE**: The text written between the [details="Summary"] (this text here) [/details] is of high value as it encapsulates the essence of the entire topic. 

# CONTEXTUAL INFORMATION ABOUT FORUM.GEORGEGAMMON.COM (REBEL CAPITALIST PRO)
- Some popular categories on forum.georgegammon.com (Rebel Capitalist Pro) are: Discussion, Rebel Capitalist Live 5, Weekly Q&As, Newsletters & Portfolios, George’s Portfolio Follow Along Journey.

# CONSTRAINTS
- When you retrieve a link, be sure that it is a link to somewhere inside this platform only.
- Do not provide any direct links that point to external websites, especially vimeo.com, regardless of the context — **unless specifically asked**.
- You must **ALWAYS** back up answers with actual search results from forum.georgegammon.com


The date now is: {time}, much has changed since you were trained.
2 Likes

I believe OP is asking about content in [details] tag. That content is fully searchable, so there is no reason why the bot shouldn’t be able to see it.

2 Likes

I’m quite sure it can’t see that either.

This could me useful too, if topics as whole are important, not only OP.

Why do you think so?

I’ve created a test post with the [details] tag and used our AI persona with Search and Read tools and it was able to search the post using the term and quote full post.

1 Like

It doesn’t see such tags, it doesn’t embed it and it can’t search it. Sure, the content is visible, but it can’t be targeted using bbcode-tags.

I have a feeling that OP wants a model reading all the content, remember it and then making a carbon copy style answer. It don’t do anything of those, plus you would be out of tokens really fast.

It can read [details] tag on single topic URLS.
iScreen Shoter - Google Chrome - 250112113201

I asked the bot to describe what the [summary] tag is and it calls it a “details” or “spoiler” block. It does much better when you are specific about calling out “details” or “spoiler” block.

No, that’s not what I want at all… I do want it to read my transcripts and newsletters contained in the “details” or “spoiler” blocks then provide simple answers to my quarries though.

1 Like

Thanks for putting this together. I’ll play around and see if it improves our results.

1 Like

If the text is in a post/topic it’s going to get passed to the LLM. It’s just a matter of how that text influences the LLMs output. This is where prompt engineering comes into play.

Yes, but in such situation you don’t need AI. Because you can go and read, but mostly because of embeds aren’t using at all that accurate context. But the example of OT, not happening.

Just? Eh… you bypassed input now totally.