Strategy explained · Polymarket crypto

Price the bet.
Buy the gap.

Polymarket asks questions like "Will Bitcoin reach $100k by September?" and lets people trade YES and NO at prices between 0¢ and $1. To an options trader each one is a digital option, and options markets already price that probability. TurtleQuant reads Bitcoin and Ether implied volatility from Deribit, turns it into a probability for every market, and buys whichever side the crowd has priced at least 5¢ too cheap.

Minimum edge
5¢
model − price, either side
Sizing
0.25× Kelly
capped at 10% of NAV per market
Holding window
7–90 days
to resolution at entry
Live order cap
$6
per order, real money
Try it

Price a market

Pick a market the bot priced in its last scan, or set your own inputs. The lab runs the bot's Black-Scholes formulas, compares the result with the order book and decides what the bot would do.

Model P(YES)
—
YES edge
—
model − YES ask
NO edge
—
(1 − model) − NO ask
—

The lab uses the bot's flat-vol ("legacy") model with a 5% rate. The live bot trades on a smile model that adds a skew correction, and it only enters when both models clear the threshold on the same side. For a scanned market the lab also applies the smile price from that scan; once you edit an input it falls back to flat-vol alone. Sizing assumes an empty book of $100.00 (the live NAV), so caps on total and per-expiry exposure don't bind here.

Every 60 seconds

How a scan runs

  1. 01 · find

    Read the board

    Pull Polymarket's crypto price events, about 1,400 markets in two requests. Keep those with over $5k of liquidity and both sides quoted within 3¢. One-sided markets are dropped, not given a made-up spread.

  2. 02 · parse

    Turn words into a contract

    "Will Ethereum dip to $2,000 by December 31?" becomes ETH, $2,000, Dec 31, touch from above. Three shapes trade: above at expiry, reach before expiry, dip before expiry.

  3. 03 · price

    Ask the options market

    Read Deribit's implied volatility for that strike and date. Price "above" as N(d₂), and "reach / dip" with the reflection principle. The smile model adds a skew term, and the flat model has to agree.

  4. 04 · buy

    Only when it's cheap

    Buy YES or NO when the model beats the executable ask by 5¢ or more, and the market resolves 7–90 days out. Size at 0.25× Kelly, inside the exposure caps.

  5. 05 · exit

    Hold, unless the bid overpays

    Default is hold to resolution. Sell early only when the bid, after Polymarket's fee, is worth more than the model's probability plus a 1¢ margin. Those show as ev exit in the trade log.

Guard rails

What stops it

Per market
10%
of NAV
Per expiry date
15%
of NAV
Total exposure
40%
of NAV
Daily loss halt
5%
of NAV, realized
Drawdown halt
15%
from peak NAV
Stale data
No entry
on old books or vol
Best shadow trade so far
Will Ethereum dip to $2,000 by December 31, 2026?

Bought YES at 17¢ on Sep 28, 2026. The model put it at 26%. Sold at 91¢ on Nov 12 when the bid paid more than the model's value.

Return
+427%
P&L
+$86.73

Rules on this page are read from the live bot's startup log each time it is built (2026-11-17 03:07 UTC). The live account is small and every order is capped; the longer record is the shadow run, which simulates fills against real order books. This describes a trading experiment and is not investment advice.