Editor's commitment

If you find any of the following in one of our articles, please email [email protected] with the subject "Correction: [article slug]". We'll reply within 72 hours, push the fix, and credit you below:

  • An OKX parameter (funding cap, MMR, contract face value, max leverage) is wrong.
  • A formula derivation skipped a step or relied on an unstated assumption.
  • A screenshot or button reference no longer matches OKX's live UI (their redesigns aren't always in our favour).
  • A numerical example doesn't compute.
  • An external citation is dead or misquoted.

Below is the full revision history, newest first. Reader credits use the pen name or email handle the reader sent us — we never publish real identities without consent.

2026 revisions

2026-09-16 · long-short-ratio-vs-top-trader.html, perp-newbie-5-mistakes.html, data/long-short-ratio.html

  • Original error: the Top Trader series was described as a cut by net PnL, trading performance or "top-ranked profitable accounts".
  • Fix: OKX's API documentation defines it as "the top 5% of traders with the largest open position value". The ranking is by open position value — not by volume traded and not by profitability — and there is no win-rate screen anywhere in it. Corrected in the article body, the FAQ (visible and structured data), the dashboard copy and the glossary entry.
  • Original error: the 100x worked example used 0.03 BTC for a notional of about $2,953, which does not follow from the contract spec.
  • Fix: a BTC-USDT perpetual contract has a face value of 0.01 BTC. The example now runs 30 contracts = 0.30 BTC, notional 30 × 0.01 × $98,420 ≈ $29,526, initial margin 29,526 ÷ 100 ≈ 295 USDT, with every downstream figure re-run from those.
  • Original error: an endpoint form that does not exist (long-short-position-ratio?ccy=) was listed, and the Top Trader endpoints were shown taking ccy.
  • Fix: the exchange-wide series takes ccy; the per-contract and Top Trader series take instId. All four endpoints are now listed with their documented limits (period lookback 2 / 30 / 180 days, 1,440 rows per call, Top Trader data from 2024-03-22, 5 requests per 2 seconds). OKX does not document a refresh interval, so the previous "refreshes every 5 minutes" claim was removed.
  • Original error: several statistics were published without a reproducible source — a long-run mean LSR band, a reversal probability after extreme readings, a hit rate for divergences, and precise LSR readings for the 2024-08-05 replay.
  • Fix: all of them removed. Thresholds are retained but labelled as common reading habits rather than official levels, and the 2024-08-05 replay is now told from roughly what the public dashboards showed at the time, with no decimal readings.
  • Original error: funding was described as a percentage of margin, on a fixed 00:00 / 08:00 / 16:00 UTC schedule.
  • Fix: funding is charged on position value (face value × contracts × mark price), and the settlement interval is set per coin — 8, 4, 2 or 1 hours, with BTC and the other majors usually on 8 hours. Converting to margin means multiplying by actual leverage, so +0.05% over three days is 0.45% of position value and about 2.25% of margin at 5x.

2026-05-12 · perp-fee-funding-guide.html

  • Original error: stated that the single-period funding rate cap on OKX BTC-USDT-SWAP was ±0.5%.
  • Fix: the cap on majors (BTC-USDT-SWAP, ETH-USDT-SWAP, etc.) per 8h settlement is actually ±0.75%, with select low-liquidity altcoins extended to ±1.5%. The relevant table and the annualised cost example were both updated.

2026-05-04 · leverage-liquidation-math.html

  • Original error: the isolated-margin liq price example dropped the MMR term when back-calculating from entry, so the printed liq distance was roughly 0.4% too tight.
  • Fix: MMR restored. A 10x BTC long that was printed as liquidating at -9.5% now correctly reads -9.1%. All downstream numbers in the piece were re-run.

2026-04-22 · grid-strategy-handbook.html

  • Original error: the geometric grid section wrote the per-grid step as (upper - lower) / n. That's the arithmetic grid step.
  • Fix: geometric grids step by (upper / lower)^(1/n). The fix flows into the ETH 9000–13000 / 100-grid budget table, which was recomputed.

Archive

The site launched in 2026-01. Internal drafts prior to 2026-02 were never published externally. The correction log starts at 2026-02 and is complete from that point onward.

How to send a correction

Email subject template: Correction: [article slug] — [one-line description]. Body should include the location (heading or a quote of the original line) and what you believe the correct version is. No need for a long email. We'll confirm by reply and credit you under the pen name or handle of your choice — anonymous credit is fine too.

Heads up: a revision log doesn't constitute investment advice. Even when our facts are right, the trade you make on top of them is still your own risk.