MEXC Both Ways PAPER · FORWARD ONLY · A/B

Both legs are open simultaneously. One will finish at −1R; the pair pays only if the other clears +1R.

Commission 13.17 bps round-trip on notional, charged on both legs. Slippage is measured, not assumed — per symbol and per size from the live depth probe (median 0.110% of price at $1,500, rising to 0.383% at $15,000). Funding is the real settlement over the hold.

★ The straddle is funding-neutral by construction — the legs are equal and opposite, so one pays exactly what the other receives. The directional FADE book pays 0.0171 R/trade in funding; this one pays ~nothing. That is a genuine structural advantage.
⚠ Every pair pays four fills (two in, two out). A pair that grosses exactly $0 — which is ~90% of them — still loses the cost. That is the whole problem this experiment is testing.

What this is. Every spike gets a LONG and a SHORT at identical size. Stops sit at ±1R and the ratchet triggers at +1R, so the moment one leg's stop is touched the other has already locked ≥+1R. A pair can never net below zero — verified 0 negative pairs of 62 across {15m, 1m} × {optimistic, pessimistic} replay, plus 76 live and 20 backfilled pairs.

Where the money comes from. Nothing, unless the winner exits above +1R. Historically only 5.3% of legs ever do. So the experiment is really one question: how often does that happen, and is it enough to pay four fills on every pair?

The two numbers that decide it.
1. Pairs below zero — must stay 0. Structural, not statistical. Non-zero means the premise is broken, not that we were unlucky. ⚠ It breaks below STOP_ATR 2.5 (1 of 69 at 2.0), because one 15m bar can take out both legs before either locks.
2. Residual rate. Historically 15% in the first half and 5% in the second — and that second half lost money with perfect fills. ≥10% over ~60 pairs ⇒ real (then spend ~$200 on live orders to measure real fills — do not fund on paper). ~5% ⇒ dead.

Why bps and not dollars. Both books size to the same risk, so TIGHT carries ~1.6× the notional — its commission runs exactly 1.60× BASE's, which is the proof. Dollars therefore contain a leverage difference. Judge the config on bps; judge a deployment on dollars.

Known limits. Paper fills are perfect, so this page can never measure real slippage on the exit stops — only live orders can. Forward-only from 2026-08-09 18:00Z; an earlier backfill was discarded because it replayed old spikes under today's ratchet while the live books had closed them under a buggy one.