I don't really think it's over but it obviously has to be improved
The other (more important) question is how to improve and would strategy still be alive after my improvements
Extended: this really is the last chancee
For a while Extended had real mindshare. Plenty of people wrote about it, some liked trading there, and some bought into the ex-Revolut team. Then the points program ended, TGE got pushed back, and a lot of people forgot about the product.
Still, while we're used to paying to trade, Extended is offering to pay us in The Final Arc competition. A fitting name, given the move off Starknet.
The race runs Oct 7 to Nov 3 and will distribute:
- $250,000 in stables across 500 winners, with bigger payouts for higher ranks
- 750,000 points split pro rata across all participants, which can be sold at roughly $0.70–0.80
Before you start trading, it's worth knowing how your activity actually gets counted.
Your place in the stablecoin pool is set by Score = PnL × Weighted Volume^0.7 × OI^0.3. Crypto taker fills count in full toward weighted volume, RWA taker fills at 50%, and maker fills at only 10%. OI is your average position size across the whole competition: hold $100,000 for one hour of a 10-hour competition and only $10,000 counts. Scores can go negative, so losing trades can drag you down the leaderboard.
Points go to every participant through Activity Score = Weighted Volume^0.7 × OI^0.3. Each trader gets 750,000 × their Activity Score ÷ the sum of all participants' Activity Scores.
Volume carries more weight than OI in both formulas. Doubling volume lifts Activity Score 1.62x, while doubling OI lifts it only 1.23x.
Fast trades don't pay here. A trader with $10M of volume who held $200,000 positions for just 2 hours in total scores about the same as one with $1M of volume and a steady $100,000 position. And since OI is averaged over the full period, a position opened now (Oct 20) counts at roughly half.
docs.extended.exchange/exten…
Lighter on Robinhood Chain plus Variational is one of the cleanest farming pairs for stocks and XAU right now.
Go long on one venue and short the same asset on the other. Price risk cancels out, and three things are left.
The pair pays you. The funding spread between the two venues runs above 20% APR on notional for these markets, so the position is net positive before a single point is counted.
It costs very little to run. Lighter charges retail accounts no trading fees and Variational's Omni charges none either, so the main cost is the spread on entry and exit.
The points stack. Every $100K of OI and weekly volume earns more than 2.5 Variational points and more than 9 Lighter Robinhood points per week.
Take $20K of margin on each leg at 5x, so each side holds a $100K position and the whole pair runs on $40K.
Over 52 weeks:
• Variational at $35 a point: 2.5 × $35 = $87.50 a week, $4,550 a year, 22.8% APR on $20K
• Lighter at $20 a point: 9 × $20 = $180 a week, $9,360 a year, 46.8% APR on $20K
Together the points pay $13,910 a year on $40K, or 34.8% APR. The 20%+ funding spread on $100K notional adds another $20,000, which is 50% on the same capital. All in, the pair returns roughly 85% APR before costs, and every step up in leverage multiplies it further.
STRC is back at par. The asymmetry now points the other way
I closed my $STRC longs at $99. Entry was $83. The thesis worked, and a trade stops being asymmetric.
Saylor used three of the four cheat codes from my last post. He set a 12% rate from July and bought back about $950M of STRC between Jul 20 and Sep 13. The buybacks were funded by selling $MSTR ($765M) and $BTC ($161M). STRC closed at $99.90 on Oct 13.
The upside is now capped. Par is where Strategy turns issuance back on and sells new STRC to buy BTC. Above par, the board can cut the rate. The buyer that pulled the price up becomes the seller sitting on top of it.
The downside has a floor and a tail. The floor is the mid-90s, where STRC sat in August even with buybacks running, about $5 below here. The tail is June. When BTC broke $60,000, STRC fell from $91.70 to a $73.80 close in a week, a 19.5% drop. A full repeat is less likely now: the 12% rate, $1B buyback program and USD reserve weren’t in place back then. But that defense is paid for by selling BTC and MSTR, and BTC sits near $83,400 with its Nasdaq correlation back at 0.6.
My math on a short from $99.90: the worst case is a squeeze to $101, a $1.10 loss. A drop back to the mid-90s, where STRC sat in August, pays $5. A June repeat pays $20+. Even the base case is 1:4, and the tail is 1:20.
Is there an ETHFI buyback at all?
Yes and no. A single Safe has bought about 1.37M ETHFI with USDC since Sep 8, but as far as we can trace, none of it stays out of circulation. Every token it buys is passed on to a reward contract on Optimism, and over Oct 5–11 that contract paid out 1.56x as much ETHFI as the Safe bought.
Tracing both wallets through Oct 12 shows four things:
1. The purchases appear to fund an ether.fi Foundation program that turns part of the protocol's fees into ETHFI. Half of what is bought was meant for the treasury and half for user rewards.
2. The buyer has spent about $957K, roughly $0.70 per ETHFI, and has not sold once.
3. The tokens do not go to a burn address or to a treasury. They go to a contract on Optimism that has sent 1.25M ETHFI to 11,221 addresses.
4. Last week the contract paid out 462,671 ETHFI against 297,400 bought. A surplus from September covers the gap, and at last week's pace it runs out in about nine days.
Why the buyback exists
The ether.fi Foundation published a proposal on Aug 30, and voting closed on Sep 3. According to Alea Research's summary, ether.fi Labs converts revenue from the Cash card, swaps and staking into USDC and sends part of it on-chain, where it buys ETHFI every week through CoW Swap. In the proposal's own example, a $50 fee splits into $20 kept by Labs and $30 used to buy ETHFI. Of that $30, $15 goes to the Foundation treasury and $15 goes to users as rewards. The proposal estimated the program at about $1.33M a month from July data and allowed up to 20M treasury ETHFI to be used if rewards ran short. On paper, half of every dollar spent on ETHFI takes tokens off the market for good, and the other half goes back to users.
Safe 0x43f4ca92… made its first purchase on Ethereum on Sep 8, five days after voting closed. Since Sep 29 it has bought on a fixed schedule: $3K of USDC every 2.4 hours, or about $30K a day.
The tokens come back out on Optimism
The same Safe address on Optimism sent nine transfers to contract 0x38f2fbb2… between Sep 8 and Oct 11: a 100 ETHFI test, then eight top-ups of up to 266K each. They total about 1.40M ETHFI, roughly 30K more than the Safe bought on Ethereum.
The contract splits those top-ups into payments with a median size of 18 ETHFI. Through 11:01 UTC on Oct 12 it made 21,838 transfers totaling 1.25M ETHFI to 11,221 addresses, according to the Optimism block explorer. Of those transfers, 1,838 went through a function named award(claimId, recipient, token, amount), and all 25 counterparties on the first page of Nansen's list for the contract are BeaconProxy contracts. That pattern fits a program paying rewards into many individual user accounts.
Regular payouts started on Sep 18: 54 transfers that day, 468 on Sep 21 and about 1,100 a day over the two weeks to Oct 11. Over Oct 5–11 the contract paid out an average of 66.1K ETHFI a day, while the Safe bought 42.5K. At daily closing prices that is about $47.7K paid out against $30.4K bought each day.
A buffer that refills every 94 hours
The reward contract never runs dry because the Safe tops it up first. The last four refills, on Sep 29, Oct 3, Oct 7 and Oct 11, came about 94 hours apart. The last three each arrived when the balance was down to 96–108K ETHFI, and each added 185–250K. As of 11:01 UTC on Oct 12 the contract held about 217K ETHFI. With no refill, that covers about three days of payouts. If the rhythm holds, the next refill is due around Oct 15.
The gap is being paid for with ETHFI bought in September. Over Sep 8–20, before payouts picked up, the Safe bought about 461K ETHFI and the contract paid out about 16K. That surplus now sits in the contract's 217K balance. If purchases and payouts stay at last week's levels and each refill only passes on new purchases, it runs out in about nine days.
What $1,000 of fees turns into
A round example shows how far the on-chain flow has drifted from the proposal. Take $1,000 of fees and assume the Safe's USDC is the 60% buyback share described in the proposal. Labs keeps $400 and $600 buys ETHFI. At last week's average purchase price of $0.716, that is about 838 ETHFI. The proposal's split would send 419 of them to users. On-chain, all 838 went to the reward contract.
Two ways to read it
The flattering read is that the program works as designed. The proposal always meant part of the purchases to go back to users, and it reserved treasury tokens for the case where rewards run ahead. Last week payouts exceeded purchases by about 165K ETHFI, or 23.6K a day. At that rate, 20M ETHFI would last more than two years. Daily recipients also more than doubled, from 554 on Sep 22 to 1,255 on Oct 11, which fits growing use of the products that earn rewards.
The same numbers support a less flattering read. The Safe spent about $957K in its first five weeks, roughly $0.83M per 30 days against the proposal's $1.33M a month. None of its purchases went to the treasury. A buyback that hands every token back out removes nothing from circulation: last week it took 297K ETHFI off the market on Ethereum and put 463K back on Optimism. If recipients sell more than 64% of what they receive, the program is a net seller of ETHFI.
So is there a buyback?
There is a purchase program, but not a buyback in the usual sense. ETHFI bought with fees reaches users within about a week, and for the past three weeks the program has paid out more than it bought, drawing down what it bought in September.
A wallet funded from Binance in September has been buying STRK perps on Hyperliquid since the L1 news. Is it driving the rally or following it?
On the data, following.
It deposited $7.5M on Sep 18 and ran a STRK long that closed at a small loss. It then sat flat until the evening of Oct 8, hours after Starknet said it is “actively considering becoming an L1” and STRK had already started moving.
Today it’s long $1.9M of STRK on $8M of equity, roughly 0.24x. That’s 3.4% of Hyperliquid open interest, and its buying is a fraction of a percent of daily volume. It’s too small to move the price.
The data can’t show who owns it or what it holds elsewhere. Our read is that it’s a momentum account that shows up after the move. The number we’d watch is funding: longs are paying about 56% APR going into the Oct 15 unlock.
hypurrscan.io/address/0xd978…
The bull case for ethereum:0x163f8c2467924be0ae7b5347228cabf260318753
AI agents are about to hit the wall bots always hit, and World ID is built to get them through it.
• Meta’s Muse and OpenAI’s dots now book travel, fill forms and buy things for consumers. Amazon has already blocked Muse from shopping, partly because it didn’t disclose it was an AI agent
• There are about 8 billion people and no limit on agents. Rate limits, sales and one-per-person allocations all depend on counting the humans behind the requests
• World ID lets a person hand a zero-knowledge Proof of Human to their agents. No name is shared, and someone running thousands of agents still counts as one human
• Okta, Vercel, Exa and Browserbase are already building it into agent tooling (chart)
• Several large US financial firms are testing a World ID deepfake check in Zoom, and TFH says companies now come to it first instead of the other way round
World names OpenAI’s dots as a fit, but there is no announced OpenAI and World integration, and OpenAI’s own dots announcement doesn’t mention World.
The bottleneck is distribution: Orbs cost several thousand dollars each and are concentrated in cities in South Korea, Japan, the US and Germany.
Agents that can’t prove a human sent them will get blocked. World sells the proof.
ethereum:0x1789e0043623282d5dcc7f213d703c6d8bafbb04 and $SCR are probably quite asymmetric here
Imagine having Consensys L2 sub $100M MCAP and ZK (even working) chain at $5M MCAP
Does Citrini Research's "Breaking The Wall" basket outperform the market?
Over the last 30 and 90 days it does, and by a wide margin, but a large part of the gap comes from a few names.
If you had bought the 15 tokens at their target weights, the basket would be up 38.2% over 30 days and 93.6% over 90 days. Over the same windows BTC gained 4.7% and 28.6%, ETH 0.3% and 38.7%.
The outperformance is broad: 11 of 15 tokens beat BTC over 30 days, and the median token is up 13.1%. Still, Derive alone added 19.3 points to the 30-day result, and without DRV the basket's return falls to 21.1%.
On diversification, 15 tickers look more spread out than they are. Group each token by where its activity lives, and 70% of the weight ends up in the Ethereum orbit:
• Ethereum L1 DeFi: 43%, Aave, Ethena, ether.fi, Uniswap, Pendle, Maple, Ondo
• Ethereum L2s and appchains: 20%, Lighter, Derive
• Base: 7%, Aerodrome
• Solana: 13%, SOL, Backpack
• Hyperliquid: 7%
• Cross-chain infra: 10%, Chainlink, LayerZero
By sector, the biggest bet is on trading venues. Derive, Lighter, Hyperliquid and Backpack together take 32% of the basket, so it is effectively long ETH beta plus perp DEX competition.
The weighting inside that bet runs against the revenue data. Hyperliquid made $73.4M in revenue over the last 30 days according to DefiLlama, which is 62.9% of the combined revenue of the nine basket protocols with comparable data. Yet HYPE carries a smaller weight than either Derive or Lighter, which get 10% each. In other words, the basket is long the challengers more than the current leader.
Valuations tell a similar story. Comparing fully diluted valuation to annualized 30-day revenue gives a wide range:
• Aerodrome: 8.8x
• Maple: 14.6x
• ether.fi: 15.0x
• Aave: 45.0x
• Uniswap: 45.9x
• Lighter: 64.7x
• Hyperliquid: 96.0x
• Derive: 98.3x
• Pendle: 115.7x
The cheapest names on this metric hold mid-sized weights, and ether.fi is the only top-weighted token with a low multiple. Derive sits near the top of the list after a 192.8% run in 30 days, so that position is a momentum trade rather than a valuation one. Revenue trends also diverge: Aerodrome and Uniswap are earning well above their trailing-year pace, while Aave's 30-day run rate sits 35% below its trailing year and Pendle's 69% below.
Supply adds a third layer of risk. Tokens with half or less of their supply in circulation make up 30% of the basket: Lighter and Backpack at 25% float, LayerZero at 40%, Ondo at 49%, Aerodrome at 50%. Future unlocks there will compete with any new demand.
So far the basket has beaten BTC and ETH, with most of the edge coming from Derive, Backpack and LayerZero over the last month. The next test is whether Derive and Lighter can grow revenue fast enough to justify their top weights before their supply catches up.
WHUF: the bull case for putting a price on trust
@ethos_network listed base:0xeeee77bc7e82c0d4166d52f58239d4c5bf41eeee on Coinbase today, and the first hours of trading put it almost exactly at the price the Foundation promised to buy it back at. That looks like a weak start, yet the setup behind it is one of the cleaner launches of 2026.
Ethos itself is an onchain reputation protocol on Base. Users vouch for each other by staking money, write reviews, slash bad actors and trade "trust" and "distrust" markets on any profile. All of it rolls up into a credibility score from 0 to 2800, which shows next to accounts on X through a browser extension and is open to other apps through an API. Faking reputation here costs real money, because a vouch can be slashed. Since September WHUF has replaced ETH in every one of these mechanics: it is the vouch collateral, the market currency, the bond for slashing and the reward token.
The first bull case is a cushioned downside. Sale buyers who keep their WHUF vouched can redeem it for 85% of the clearing price, $3.57, for 12 months. The token trades at $3.58–3.66, close to that floor, and any buyer who wants the protection can't sell without losing it, so part of the float stays locked by design.
How the sale was run is the second argument. Instead of selling 2% of supply to a handful of funds, Ethos auctioned 20% on Sonar, Echo's public platform, at one clearing price for everyone:
• Clearing price: $4.20, about $42M FDV
• Raised: $8.4M from 1,560 participants
• VC exposure: about 1.1% of supply, no outside investor above 1%
Ethos calls it the third-largest public sale of 2026, while the median sale this year offered only 2% of supply.
Supply is the third point. Max supply is fixed at 10M with no mint function, and every non-zero protocol fee will be burned. About 2.34M WHUF became liquid today, and the next scheduled unlock only comes around March 2027, when early supporters get 276.5K. Team tokens sit behind a one-year cliff until September 2027.
Valuation adds a fourth. At roughly $36M FDV, WHUF is cheaper than the reputation and identity tokens it gets compared to (prices as of October 7–8):
Intuition (TRUST), the only peer with a similar stake-on-claims design: $53M
• Gravity (G): $55M
• Billions (BILL): $100M
• Kaito (KAITO): $298M
• Humanity (H): $698M
The design also works against whales. A WHUF vouch counts by the voucher's rank among all vouchers, not by the amount staked, so buying more tokens barely buys more credibility. A score that is hard to game is exactly what an outside app would want to plug in, and some already do: the HCS-25 agent registry standard uses Ethos scores as a reputation signal for AI agents.
The case has clear holes. Ethos discloses no revenue, protocol TVL sits around $1M, and the guarantee expires in October 2027, a month after the team and treasury cliff.
So far Ethos has sold a fifth of its supply at $42M FDV and opened trading right next to its own put. The next twelve months will show whether wallets, marketplaces and agent registries need the score enough to lock WHUF for it.
