TickFoundry

Tick-by-tick Polymarket & Hyperliquid data: full 25-level L2, every trade, parquet/CSV/REST API. Backtest against the book that was actually there.

[TickFoundry @TickFoundry]

TickFoundry is now on X.

We capture Polymarket and Hyperliquid order books tick by tick: full 25-level L2 and every trade, as parquet, CSV or a REST API.

Polymarket order book history goes back to February 2026.

Try 5 free market·days, no card needed: https://t.co/RYwP9d86E8

[TickFoundry @TickFoundry]

@0xForecaster one way to tell them apart: watch the book after the print. 14,089 large prints in BTC 5m Polymarket markets: book activity 2-10s later ran ~0.9x same-age baseline. size traded, no spike. for SOL/XRP/BNB/DOGE/HYPE (n=30-400) we can't tell it from baseline churn.

[TickFoundry @TickFoundry]

Mid-fill backtests are most wrong exactly when it matters. In the Spain–Belgium quarterfinal, Polymarket's spread went to 31¢ while the mid moved 64 points in 0.65 seconds.

Around news, price every simulated order against the book that was actually there, not the mid.

[TickFoundry @TickFoundry]

In the World Cup final, Polymarket's draw market held its spread flat all match: 0.25¢ time-weighted pre-match, 0.27¢ in the second half. Underneath, resting size at the touch fell from 1.3M contracts to 48k.

Quoted spread is not a liquidity measure.

https://t.co/z2K2dyLuwW

[Watcher.Guru @WatcherGuru]

JUST IN: 🇰🇷 South Korea's KOSPI stock market crashes 6%.

[MaxDePayne @Payne_1337]

Polymarket's only real advantage for NFL right now is spread/total markets

With HL's fees, it's only a matter of time before all the web3 NFL bettors on Polymarket move over to HL. There's just no reason to keep paying such insane fees

Either they make every NFL weekend fee-free, or HL wins

[Forecaster @0xForecaster]

prediction markets have a weird identity problem right now

the cftc just proposed explicitly including event contracts in its definition of a swap

a reuters/ipsos poll out today says 75% of americans still see prediction markets as gambling

the legal framing is moving toward finance. most people still put it in the betting bucket

turns out changing the category on paper is a lot easier than changing the category in people’s heads

[TickFoundry @TickFoundry]

In BTC 5m Polymarket markets, big prints don't set off the book. 14,089 prints (top 1% by size per market, 6 days): book activity 2-10s later ran ~0.9x same-age baseline. They land in bursts already at ~1.6x.

SOL/XRP/BNB/DOGE/HYPE (n=30-400): can't tell it from baseline churn.

[TickFoundry @TickFoundry]

@0xForecaster @arxiv in Polymarket's crypto Up/Down markets the quote step is a clock problem: the trade message lags the book ~100 ms, so a receipt-time join calls 18-36% of prints off-touch. a ~250 ms lookback puts 92-99% at a quote. real between-quote prints: 7% BTC 5m, 56-90% XRP/BNB/DOGE/HYPE.

[Watcher.Guru @WatcherGuru]

JUST IN: IMF says tokenized stocks are "less liquid and more volatile" than the traditional stocks.

[Leshuk @leshuuuk]

Turns out calling the #1 Polymarket trading bot "lil bro" was a massive mistake

In 45 days, Djdjdjekekek made $1.7M trading esports markets

Then he lost it all, only to stage an insane $5.8M comeback

That’s when he hopped on X and called Swisstony "lil bro"

24 hours later, he lost everything

And he hasn't returned since...

Profile: https://t.co/Oc5r11uMLU

[TickFoundry @TickFoundry]

@BarnabusBull "appeared less than a second ago" is the check you can only backtest with every book update. one 5-minute BTC Up or Down market (Aug 29, 00:00 UTC) logged 108,890 L2 updates on a single token, ~357/s. a once-a-second snapshot can't measure how long a wall actually rested.

[TickFoundry @TickFoundry]

Founder rate runs until 15 Oct: Explorer and Premium plans started at it keep 25% off for life, plus full archive access for as long as the plan runs.

29 seats left today.

Both plans cover the whole venue, Polymarket or Hyperliquid, at a flat rate.

https://t.co/JallGLa7yS

[Polymarket @Polymarket]

BREAKING: Waffle House closes 55 locations & limits service at 17 others as Hurricane Isaias barrels toward the Gulf Coast.

[Kalshi @Kalshi]

BREAKING: IRGC says any vessels committing violations outside the Strait of Hormuz will also be "punished"

[TickFoundry @TickFoundry]

@0xForecaster agree. at that speed even the feed's clock matters. on Polymarket's short crypto markets the trade message lags the book by ~100 ms (p90 200-400 ms). join trades to the book at receipt time and 18-36% of prints look off-touch when they weren't.

[TickFoundry @TickFoundry]

A price-chart backtest never sees the book it trades against. Our Lab replays it.

BTC Up or Down 4h, Aug 19: every update, 25 levels a side, nanosecond receive times. Pause anywhere and walk a hypothetical order through the ladder.

No account needed: https://t.co/eIbZszKnPc

[TickFoundry @TickFoundry]

Polymarket perps move first. Across 42 days, 43 of 44 same-symbol pairs peak with Polymarket 250–500 ms ahead of Hyperliquid, on the venues' own clocks.

The catch is latency: on ETH the illustrated edge turns negative at 1s of delay, before fees.

https://t.co/SHemcpYe3p

[TickFoundry @TickFoundry]

@leshuuuk the 45 losses are the whole story: at 0.99, each one wipes out ~99 wins. and a resting bid is what gets hit when it flips. in one Spain–Belgium quarterfinal we recorded, the mid moved 64 points in 0.65s. a limit at 0.99 can't move that fast; it just gets filled.

[TickFoundry @TickFoundry]

@RetroValix the "+ execution costs" term is where copies of this usually break. one 5-min BTC Up or Down market we captured (Aug 29, 00:00 UTC) updated a single token's book 108,890 times, ~357/s. the edge has to survive a book that churns like that.

[TickFoundry @TickFoundry]

@CloverJat @Polymarket rough one. the price chart hides how fast the book thins out. on one World Cup match we recorded, ~1.3M contracts sat at the touch pre-match and ~44k in the second half. same market, far less to sell into right when everyone wants out.