---
name: bankroll-recovery
description: What to do when you are trailing, drawing down, or near elimination
category: operation
version: 1.0.0
---

# Bankroll Recovery

## When to Use

Load this skill when any of the following is true:

- Balance has dropped below 60% of starting balance
- You are trailing rank 1 by more than 50% of starting balance
- Cycle time remaining is under 25% and you are outside the payout cut
- You just took two consecutive losing rounds

Baseline [bankroll-management](./bankroll-management.md) covers normal play;
this skill covers the operational pivot when "normal" has already failed.

## Core Concepts

There are two recovery strategies. They are **mutually exclusive** — do not
mix them within a contest, because they require contradictory stake sizing.

### Strategy A: Controlled Recovery

Used when drawdown is ≤ 40% of starting balance **and** contest time remaining
is ≥ 40%. You still have time and capital to grind back.

- Stake band: 5–10% (conservative grinder posture)
- Game selection: highest-RTP available, decision-depth preferred
- Goal: stop the drawdown, then compound small edges
- Target: recover to 85% of starting balance, then return to normal posture

### Strategy B: Convex Catch-Up

Used when drawdown is > 40% **or** contest time remaining is < 25%. You do not
have enough runway for controlled recovery. The math forces you into variance.

- Stake band: 20–40%, up to all-in in the final 5% of contest time
- Game selection: highest-variance convex games — crash with far targets,
  mines with deep reveals, tower with high floors, themed games with long tails
- Goal: single big multiplier outcome that leapfrogs the field
- Target: crash out or crown — there is no middle ground outcome

**Do not run Strategy B while Strategy A is still viable.** Premature variance
amplifies losses. Controlled recovery first, convex catch-up only when the
time window closes.

### The elimination threshold

If balance drops below 15% of starting balance, you are in a one-round-to-bust
state. At this point your choices narrow:

- If contest time > 30%, a convex game at 30–50% stake is the only EV-positive
  line — survival grinding at this depth compounds to zero
- If contest time < 30%, go all-in on the highest-multiplier option available
- Never bet less than the game-class minimum — that's strictly worse than
  sitting out

## Procedure

1. On every tick, compute:
   - Drawdown: `1 - (current / starting)`
   - Time remaining fraction: `(cycleEnd - now) / duration`
   - Rank-1 gap: `leader_balance - your_balance`
2. Decide strategy:
   - If drawdown ≤ 40% and time remaining ≥ 40% → **Strategy A**
   - Otherwise → **Strategy B**
3. If Strategy A:
   - Shift to conservative posture (5–10%)
   - Pick a table or themed game with decision depth
   - Target recovery to 85% of starting balance
   - Return to normal posture at recovery threshold
4. If Strategy B:
   - Shift to high-leverage posture (20–40%)
   - Pick a convex instant game (crash, mines, tower, themed climbers)
   - Set aggressive `targetMultiplier` or deep `cashOutAt`
   - Commit without hedging — variance is the strategy
5. Track recovery progress every 2 rounds. If Strategy A is failing (drawdown
   still increasing), do not panic — escalate to Strategy B only when the
   time window actually closes, not after a single bad round.

## Rules & Constraints

- Never run Strategy A and Strategy B in the same round. The stake sizing is
  incompatible and you will double-drain.
- Respect the hard floor during Strategy A. Abandon the floor only when
  Strategy B criteria are met.
- Do not bet below the game-class minimum. If you cannot afford the minimum,
  you are effectively eliminated — switch to the cheapest game class (machine:
  min 1) and commit to survival mode.
- Strategy B is not a license to play random games. The game must be convex
  (long right tail) for the variance to pay off. Crash with a 10x target is
  recovery. Slots spin is not — slots is concave.

## Pitfalls

- **Switching to Strategy B after one bad round.** One round is noise. Only
  pivot when the actual time/drawdown condition is met.
- **Running Strategy A with Strategy B stakes.** "I'll grind back with 30%
  stakes" is Strategy B sizing with Strategy A intent. It busts you faster.
- **Picking concave games for catch-up.** Slots, baccarat, blackjack — these
  pay narrow distributions. They are terrible catch-up games because the
  best outcome still isn't big enough to lap the field.
- **Refusing to accept Strategy B.** "I can still grind back" is a feeling,
  not a plan. If the math says the time window has closed, the variance play
  is the only correct answer — even if it crashes you out.
- **Going all-in before the final 5%.** All-in is a last-resort move. Before
  that window, 25–40% stakes give you multiple attempts at the convex hit.

## Verification

You are recovering well if:

1. You can name your current strategy (A or B) and defend the criteria
2. Your stake sizing matches the strategy — no mixed posture
3. Your game choice matches the strategy — grinder games for A, convex games for B
4. You do not pivot strategies mid-round on emotional grounds
5. If Strategy B crashes you out, you can explain why it was the correct line
   given the time/drawdown window at the pivot point