When not to exploit

In this lesson

• Judge whether a sample is big enough to act on.
• Recognise opponents who adjust and exploit you back.
• Weigh what an exploit gains against the cost of being wrong.
• Fall back to the baseline when the read is unclear.

Read first: Node-locking in Rail: the best response to a read, Reading opponent stats, and when to trust them · Advice, not a lock: every lesson opens.

Concepts used: GTO and exploitative play, EV, EV loss and how Rail grades

Too few chances

An exploit is a bet on a read, and a read on ten chances is mostly noise. Before any adjustment, put the interval on the stat and ask whether all of it sits on one side of the line that decides.

The counter costs more than the gain

In the lab's hand-worked game (lesson X3-L1) the read gains 1.25 and a caller who notices and adjusts takes 2.5 back. On real rivers the lab finds the same shape on almost every river it tried: what an exploit loses to a counter is larger than what it gains, while the baseline's own loss to its counter is small.

So an exploit pays only if the opponent is unlikely to adjust before it has paid. With those numbers, if the chance he counters is a, the read is worth (1 − a) × 1.25 − a × 2.5, which breaks even at a = 1.25 / (1.25 + 2.5), one third. Against a regular who watches you, a third is not much.

When the read is wrong

The other risk is a read that was never right. In the toy game a wrong read costs nothing, because the baseline caller is exactly indifferent. On real rivers the lab's best response usually loses less against the baseline opponent than it gains against the locked one — but rarely nothing, and more the further the lock is from the solve.

Fall back to the baseline when the sample is thin, when the player adjusts, or when a wrong read would cost a lot. Exploit freely the players who do not adjust, on reads your own numbers support.

From Rail's exploit lab: the read locked on a river solved on demand, the rest of the opponent's strategy kept at the solve; the counter is the opponent's best response to your exploit. The ranges rest on Rail's narrowing model.

Rules of thumb

• No interval clear of the line, no exploit.
• Expect a counter to cost more than the exploit gains.
• Exploit hardest the players who do not adjust.
• When in doubt, the baseline.

When they break

• Against a player who never adjusts, the counter never comes: lean on a good read.
• Some adjustments cost almost nothing if wrong: those need less certainty.

Example hands

Your own pool

Practice

The lesson is mastered when every exercise below that Rail can grade is passed. Missed items come back later as review cards.