August was the strongest month for crypto prices since spring. It was also the month of our 23 August incident. The second part has its own report — the post-mortem covers what happened, why, and what we changed, and we are not going to retell it here. This review covers what shipped in August and one question we have been asked more than any other since the month closed: how a market that went up 20–40% produced negative coin results in every Alpha Vault except USDT.
Product updates
PARK on Uniswap
PARK now trades on-chain. We launched a Uniswap v4 pool on BNB Chain, available directly through swap.earnpark.com: zero pool fee and on-chain limit orders, so you can set a price and let the order fill without watching the chart. Until now the only venues were centralised exchanges; this adds a route that does not depend on any of them.
Smaller changes
- Updated landing page.
- UI fixes on the PARK token page.
- Intercom issue fix.
- Technical improvements and performance optimisation across the platform.
Strategy performance — August 2026
Earn strategies
Maker Core, BTC LP, ETH LP and ETH DeFi were affected by the 23 August liquidation; USDT and USDC DeFi had a separate result the same week. Both are covered in full in the post-mortem, including the impact by strategy and every change that followed. We are not repeating it in this report. The remaining Earn strategies accrued and paid out as usual; current rates are on the strategy pages.
Alpha Vault strategies (Advanced Yield – High Risk)
Alpha Vault results are reported in the vault's own asset. A BTC vault result is a change in the number of BTC in the position; a USDT vault result is a change in USDT. This matters for reading August.
One vault in plus, six in minus, and the deepest minus on the asset that rallied the hardest. In July SOL was the only coin vault in positive territory; in August it sits at the bottom of the table. That is not a coincidence, and it is not six separate failures. It is one mechanism, and August showed it at full strength.
What the market did. Bitcoin went from the low $60,000s to above $80,000 in August, Ethereum added roughly 30%, Solana roughly 40%. Almost none of that happened gradually. For the first two and a half weeks most of the line-up traded flat, with small dips into mid-month. Then, from 19 August, a US Treasury announcement set off one of the largest short-liquidation waves on record — about $2.8 billion in short positions closed in 24 hours, roughly $1.3 billion of it inside a single hour — and the market added nearly all of its monthly gain in one week. The last week was a shallow correction, a few percent off the highs. The month's shape, then: quiet, one steep week of forced buying, quiet again.
What the vault did with it. Each Alpha Vault holds roughly half of its position in the asset and half in stablecoin, and earns from trading fees and liquidations on the venue. Every time the price moves, the two halves drift out of proportion and the vault rebalances back to 50/50. On a rise, the asset half grows past 50%, so the vault sells some of the asset to restore the balance. Each further step up sells a little more. Over a month in which the price went up, the vault sold asset all month long — to the same buyers who were pushing the price. By 31 August the position held fewer coins than on 1 August. Since the result is measured in coins, fewer coins is a minus.
Nothing left the position. Every coin sold became stablecoin, and that is exactly what the USDT vault result shows: +12.47% for the month is the other half of the same trade. The coin vaults booked the asset side of the rebalancing; the USDT vault booked the stablecoin side.The shape of the move decided how much fees and liquidations added on top. Rebalancing on its own is indifferent to the path — it only cares where the price started and where it ended. What the vault earns above that comes from two-way movement: a rally that retraces gives it liquidations to collect and asset to buy back lower, and the more of that a month has, the more the vault adds to the mechanical result. The first half moved in a narrow range. The one large move went one way, in a week, on forced buying — the vault sold into it and had nothing to buy back until it was over. The correction that followed was a few percent, enough to buy back a fraction of what had been sold, not more. Compared with the two-way swings of early summer, there was simply less for the vault to work with — which is why the results sit on the benchmark rather than above it.
How deep the minus should have been. The fair comparison is not with holding the asset, which is a different product with a different risk profile. It is with the mechanism itself: a portfolio that holds 50% asset and 50% stablecoin, rebalances back to that split continuously, and earns nothing from fees or liquidations. That is the mechanical benchmark — what the rebalancing alone would do to the coin count, with no strategy on top.
For a month of significant growth the benchmark is simple: the stronger the rise, the more asset is sold, and the deeper the coin result. A 10% rise takes the coin count down by about 5%. A 21% rise takes it down by about 9%; a 31% rise, by about 13%; a 40% rise, by about 15%.
In general the vaults landed within about one percentage point of the benchmark, on both sides of it. That is the answer to the August question. The minus is the rebalancing, sized by the rally: the depth of each vault's coin result tracks the strength of its asset's move almost exactly, which is why the spread between AVAX and SOL in the table mirrors the spread between their price moves. What fees and liquidations contributed on top was small in a month where the price never came back — the vault sells into a rally and earns most from two-way movement, and August had very little of that.
For the record, and only for the record: in dollar terms every Alpha Vault ended August higher than it started. The result is a coin count, not a loss of value. We report in the asset because that is what the vault holds and what you deposited, and we will keep reporting that way.
The same mechanism in reverse. In a falling month the rebalancing buys asset as the price drops — the stablecoin half grows past 50%, the vault buys the asset to restore the balance, and the coin count goes up. That is where the positive coin results of past months came from: the vault accumulates when others are selling and returns asset to the market when others are buying. The July report had SOL as the only coin vault in plus; in August it is the deepest minus, on the steepest rally in the table. The mechanism did not change between the two months. The direction of the price did.
Closing
August is the clearest illustration we have had of what an Alpha Vault result means. It is not a bet on price, and it should not be read like one. A strong one-directional rally is the one market where the coin result and the price move point in opposite directions, and August was that market from the 19th onwards. The vaults did what the mechanism says they do, to within a point of the benchmark, and they hold what they held in August: a rebalanced position, with more of it in stablecoin than a month ago.