Memecoin Mania & Biden's Blunder

Happy Friday and welcome back to the LO:DOWN. We took a brief hiatus during the end of summer break in the UK and over Labour Day weekend in the U.S. and there’s a lot to catch up on since then. Last time we spoke Scott Bessent had just announced his bond buying program and BTC had its biggest weekly Dollar gain ever as a result. Prices have since cooled, as has the impact of Bessent plans as he continues to grapple with high long-end yields with the 30-year at a 24 year high above 5.3%.

Despite the rally losing momentum crypto majors are all still above August 19 levels, with HYPE and XRP the big winners notching returns of around 35%. This leaves crypto ahead of every other asset class on an absolute returns basis, although WTI and Brent both come close with double-digit returns (15% and 12%, respectively) of their own over the period.

Top of book depth across the six majors is about 30% thinner than it was before August 19th, and nearly a month on there is no sign of it filling back in. Bitcoin carries most of that fall, with the median dollars resting at the best bid and offer running $472k a day before the break against $300k since, with the latest day back up at $311k.

Solana and XRP lost a similar share at 38% and 40% on the same before and after medians, while Ethereum and BNB came through almost untouched at 6% and 8%. HYPE went the other way and roughly doubled from a $6.9k median to $13.2k, which says less about Hyperliquid winning flow outright than about where the market makers who pulled quotes elsewhere still wanted inventory.

These are quotes rather than fills, so a thinner top of book can coexist with unchanged execution if the resting size has simply moved deeper into the ladder. It still changes how I size, because a book showing a third less at the touch punishes market orders first, and any hedged position depends on getting two legs on within seconds of each other without paying up on either.

The date the depth broke is the part worth sitting with, because August 19th is when Bessent announced the intervention plans, and the fortnight that followed did not resolve into anything a market maker could price. Kevin Warsh then used his first Jackson Hole speech as chair on August 28th to sharpen the inflation warning, and the market read it as an endorsement of tightening.

So the September meeting on the sixteenth now prices a quarter point increase to 3.75% to 4.00% at roughly 60%, on a rate that has not moved since December 2025. Treasury signalling support and the Fed signalling restraint inside ten days of each other is an unusually wide split, and it is a reasonable explanation for why nobody wants to warehouse inventory at the touch.

I want to be careful here, because a plausible story about why depth fell is not evidence that it predicts anything, and I have no out of sample work behind it. What I would say is that the explanation is testable, and if it is right then depth should rebuild once the path is settled rather than on any particular outcome. There are three meetings left in the year, on September 16th, October 28th and December 9th, so the rest of 2026 runs inside that repricing.

Robinhood Chain has been live since the end of June, and the headline number attached to it is $17.7bn of pool volume. That sounds like a serious market until you look at where the volume actually sits.

Almost $15bn of the total is four pools pairing USDG against WETH, which is routing plumbing rather than a market anybody is expressing a view in. Take those out and the honest figure for real trading is $2.73bn, split fairly evenly between memecoins and the stock-paired names.

That is a respectable number for a chain ten weeks old, and it is also about a sixth of what the headline implies, which is the sort of gap worth checking before anyone writes that a new chain is doing billions.

The memecoin half of it is really one name, because PONS accounts for $1.07bn of the $1.43bn on its own, with JUGGERNAUT and FRONG splitting most of what is left and a thin tail sitting underneath them.

Both PONS pools peaked on September 6th and have roughly halved since. Chain-wide meme volume did set its high four days later, but that print is JUGGERNAUT and FRONG jumping twentyfold for a single day before falling straight back, which reads as one afternoon of rotation rather than a trend taking hold.

Set all that against the broader memecoin market, which is where the popular version of this story goes wrong. The meme complex did get busier, but it happened on August 19th rather than in September, and it happened alongside everything else that moved that day. Daily turnover excluding USELESS averaged $557m over the first eighteen days of August and $2.5bn over the fortnight that followed.

September has settled at $1.5bn a day, comfortably above the quiet stretch before the break and comfortably below the panic right after it. Calling that a memecoin revival in September gets the date wrong by about two weeks.

The one genuinely new thing in September is USELESS, which went from $4m a day over the first eighteen days of August to $387m a day in September, a factor of 88. Every other name is trading at a level it first reached three weeks earlier.

Which brings us to Hunter Biden's LAPTOP, a memecoin designed by former President Biden’s son and meant to poke fun at his past indiscretions in a self deprecating way. This isn't the first politically-linked memecoin, it sadly won't be the last, and it arrived late.

It listed on September 9th, printed $5.40, and traded $62.9m across Gate, KuCoin and Kraken in the 46 hours that followed. By the next morning it was the eleventh largest memecoin by turnover, above both FLOKI and POPCAT.

By the morning after that it was changing hands at $0.44, down 92% from the first print, with the hourly volume decaying inside a day and a half rather than settling into any kind of range.

None of that is a failure of execution, because median quoted spreads ran between 30 and 45 basis points across the three venues, and Kraken's wild opening hour was back inside 50 basis points by the next one.

What was missing was anywhere for the flow to come from. The attention a launch like this needs had already been allocated, first in the August 19th re-rating, then into USELESS from the end of the month, and then into PONS and the other chain-native names whose own peak came three days before LAPTOP existed.

So Robinhood Chain has a genuine market, small and very concentrated, while the complex around it has been running at the same level since late August rather than building, and LAPTOP launched into a window that had already closed. A listing with good spreads and no incremental buyer is the exact trade that looks liquid on the way in and costs you on the way out.

Three central banks inside 48 hours, and the Fed is the one with the genuine two-way risk. A quarter point rise to 3.75% to 4.00% is priced at roughly 60%, so the market is not positioned for either outcome with any confidence, and the Wednesday afternoon print lands in the same session as August retail sales.

The Bank of England is expected to hold at 3.75% on Thursday, with UK labour market data on the Tuesday and August CPI on the Wednesday morning giving the committee two chances to be wrongfooted before it votes. The Bank of Japan is the live one nobody is watching, since a rise to 1.25% has moved from possible to likely.

Worth flagging that crypto trades straight through all of this, while anything you hold on trade.xyz is pricing off an oracle that tracks a venue which is shut at the time, and the book was thin going in anyway.

Bank Rate is 3.75%, the fed funds target range 3.50% to 3.75%, and the BoJ policy rate 1.00%. The Bank of England does not meet again until 5 November, so the October cluster is a Fed and BoJ week.

Now, time for something new…

Starting Monday September 21st I am putting real capital to work, and running an AI agent with its own wallet alongside me on the same universe of Hyperliquid markets. Same start date, same end date of 14th of December, same rules on drawdown, and both books reported in full every week.

I've read and seen a lot being said about AI-enabled trading lately, such as the YC firm claiming a model they trained notched up 120% returns as the broader market was flat. So, I want to test out how good AI can be for myself, but more than that I'm testing my knowledge too. I have spent years writing about how other people trade, covering liquidity conditions, derivatives, risk metrics and index construction without ever having managed risk myself.

Every trade from both books will be published with timestamps and on-chain records, so you can check the arithmetic rather than take my word for it. Any discretionary positions come with a pre-trade note setting out the thesis and the invalidation level, written before the fill rather than after it, and the agent's prompts and reasoning go out the same way.

If you want to watch a first-time risk manager find out what he actually knows, and find out at the same time whether a machine can do better or worse, the first issue of McMargin Called will go out to all LO:DOWN subscribers.