Know when the market is too fragile to trust.
Tachyon reads the Bybit order book for BTC, ETH and SOL perpetuals and gives you one of four states. It tells you when conditions are unsafe, not which way price will go.
Four states, your rules
Tachyon publishes the state. What you do about it stays your decision.
NORMAL
Nothing unusual in the order book.
Trade as you planned.
SHIFT
Activity is well above this pair's recent range.
Widen spreads or reduce size.
ANOMALY
Unconfirmed stress: a sharp move or a spread blowout.
Pause new entries for a moment.
CIRCUIT_BREAKER
Confirmed stress or a liquidity vacuum.
Stop entering. A bot can flatten.
The everyday value
On calm-looking prices, the next 30 seconds have 1.3 to 1.7 times the normal spread of outcomes when Tachyon flags the market. That is a reason to change size, leverage or stops.
| NORMAL | 19.7% |
|---|---|
| SHIFT | 28.9% |
| ANOMALY | 35.7% |
| CIRCUIT_BREAKER | 38.1% |
The bad days
Warning before each crash in replay. In all nine cases an account that followed the state was saved, and no breaker fired in the quiet test hour. See one replayed.
| Crash | BTC | ETH | SOL |
|---|---|---|---|
| 2024-04-13 | 301 s | 75 s | 14 s |
| 2024-08-05 | 967 s | 358 s | 1,218 s |
| 2025-10-10 | 105 s | 1,211 s | 390 s |
What we do not claim
- Average fill cost is not worse in a flagged state. The value is in the tail, not the average.
- A plain volatility filter does about as well overall. Our edge over it is on calm-looking prices, and it is small.
- The crash results are not blind: those crashes were used to choose the rules. They are three crashes on one venue.
- The bad-fill and spread figures are measured on BTC only.
- On days of large moves the breaker can be on for up to 7% of the day.
- It does not detect spoofing, and it does not predict direction.