Martingale, Fibonacci, and D’Alembert are position-sizing progressions that try to recover prior losses by changing the next stake. The best choice depends on your bankroll, tolerated losing streak length, and how strictly you can follow stop rules. In Thai trading circles, this is the core of เปรียบเทียบ มาร์ติงเกล ฟีโบนัชชี ดาเลมแบร์ for real execution.
Strategy snapshot: risk, reward and operational complexity
- Martingale: fastest recovery when a win arrives, but highest tail-risk; capital demand grows exponentially.
- Fibonacci: smoother escalation than Martingale; recovery is slower and not always complete after one win.
- D’Alembert: smallest step changes; more forgiving operationally, but can lag in strong mean-reversion breaks.
- Conservative persona: usually fits D’Alembert or capped Fibonacci with strict loss limits.
- Aggressive persona: may choose Martingale only with hard caps and low base fraction.
- Algorithmic persona: benefits most from Fibonacci/D’Alembert due to clearer rule automation and drawdown control.
Martingale mechanics: progression, probability and ruin risk
มาร์ติงเกล คืออะไร in practice: you increase stake after a loss (classic: double) so that the next win recovers prior losses plus a small profit. The selection criteria below help decide if Martingale belongs in your toolset at all.
- Max losing streak you can survive: required stake after n losses is B×2^n (classic). If you cannot fund that many steps, ruin risk is structural, not psychological.
- Cap discipline: a Martingale without a pre-defined max step (nmax) is not a strategy; it is a hope-based escalation.
- Market regime fit: works best in high hit-rate, low-volatility, mean-reverting setups; performs poorly when a trend extends beyond your step cap.
- Execution latency and slippage: doubling magnifies costs; spreads/fees can turn a mathematically neat progression into a negative recovery cycle.
- Risk per cycle: define the maximum loss per full sequence (from step 0 to step nmax) as a fixed % of bankroll, not as “whatever happens”.
- Asset constraints: leverage limits, minimum order sizes, and margin requirements can prevent placing the next required step exactly when needed.
- Emotional tolerance: if you routinely override rules at step 3+ (panic reduction or revenge increase), Martingale is operationally incompatible.
- Persona guidance:
- Conservative: avoid classic doubling; if used, choose tiny base B, low nmax, and treat it as an experiment.
- Aggressive: only with explicit cycle stop-loss and a plan for “what next” after a failed cycle.
- Algorithmic: implement with hard constraints (max steps, max exposure, timeouts) and logging; never allow uncapped escalation.
Fibonacci system: sequence-based sizing and recovery limits
กลยุทธ์ ฟีโบนัชชี เทรด typically sizes stakes using 1, 1, 2, 3, 5, 8, ... after losses. After a win, many variants step back 1–2 positions rather than resetting fully. Compared to Martingale, growth is slower; compared to D’Alembert, jumps can still become large in long streaks.
| Variant | Who it fits | Pros | Cons | When to choose |
|---|---|---|---|---|
| Classic Fibonacci (advance on loss, step back 2 on win) | Intermediate discretionary traders | Clear rules; smoother than doubling; partial recovery dynamics are intuitive | May need multiple wins to fully recover; large stakes after long streaks | When you want structured escalation but can’t tolerate Martingale jumps |
| Reset-to-start on win | Conservative persona | Simple bookkeeping; reduces overexposure after a recovery win | Slower recovery if wins/losses alternate; can churn | When your priority is exposure control over fastest recovery |
| Step back 1 on win (instead of 2) | Aggressive persona | Recovers faster than step-back-2; stays “engaged” in recovery | Higher average exposure; can linger at higher steps | When your setup has stable win-rate and you accept higher variance |
| Capped Fibonacci (max index, then stop the cycle) | Algorithmic persona; risk-managed traders | Defines worst-case; easy to backtest; prevents tail blowups | Accepts realized losses when cap hit; requires re-entry plan | When you must guarantee a maximum drawdown per cycle |
| Fractional Fibonacci (multiply sequence by a base fraction of bankroll) | Conservative & algorithmic personas | Scales with capital; adapts across instruments with different tick values | Needs frequent bankroll updates; sizing errors if not automated | When you trade multiple assets and need consistent risk normalization |
| Hybrid: Fibonacci for losses + fixed take-profit/stop-loss per step | Discretionary traders with strong edge | Prevents “let it run” drift; ties progression to a real trade plan | Can cut recoveries short in choppy markets; more parameters | When you already have a tested entry/exit and only need sizing rules |
D'Alembert approach: incremental adjustments and volatility fit
ระบบ ดาเลมแบร์ คืออะไร: increase stake by a fixed unit after a loss and decrease by a fixed unit after a win. It is linear, not exponential, so it tends to be more bankroll-friendly but less “catch-up” aggressive.
- If your strategy has modest edge but experiences medium losing streaks, then D’Alembert with small unit steps can keep drawdown survivable.
- If volatility expands (wider ranges, more stop-outs), then reduce the unit size or add a volatility filter that pauses escalation during high ATR regimes.
- If your wins and losses alternate frequently, then D’Alembert often behaves more smoothly than Fibonacci (less jumpiness) and reduces churn stress.
- If you trade strong mean-reversion only during specific sessions (common in TH time-zone scheduling), then confine D’Alembert cycles to those windows and reset outside them.
- Persona guidance:
- Conservative: choose D’Alembert with a low unit and a strict daily loss cap.
- Aggressive: D’Alembert can be too slow; consider Fibonacci step-back-1 instead, but keep a cap.
- Algorithmic: D’Alembert is easiest to automate and monitor; add safeguards (max steps, max exposure, cooldown).
Capital planning: required bankroll, max drawdown and failure scenarios
- Define your unit risk as a bankroll fraction (example formula: unit = bankroll × r) and choose r you can keep consistent under stress.
- Pick a hard cap on progression steps (nmax) and treat it as non-negotiable; the cap is your survival rule.
- Compute max exposure per cycle:
- Martingale classic: total staked over n losses is B(2^{n+1}-1).
- Fibonacci: total is B × (F1+...+Fk) (sum of used steps).
- D’Alembert: total grows roughly like arithmetic series as steps increase.
- Stress-test a failure scenario: assume you hit nmax losses in a row; confirm margin/position limits still allow every planned step up to the cap.
- Set stop rules at three levels: per trade (SL/TP), per cycle (max steps or max loss), and per day/week (circuit breaker).
- Choose the progression based on what breaks first:
- If capital constraints break first, prefer D’Alembert or capped Fibonacci.
- If operational discipline breaks first, prefer simpler rules (D’Alembert or reset-on-win Fibonacci).
- If recovery speed is your main requirement and you can cap strictly, consider a constrained Martingale.
Persona-fit matrix: which strategy suits which trader profile
These are the most common selection errors when people try to เลือกกลยุทธ์ มาร์ติงเกล ฟีโบนัชชี ดาเลมแบร์ ให้เหมาะกับทุน but end up choosing based on emotion.
- Choosing Martingale because it “wins back fast” while ignoring that step growth is exponential and capped by reality (margin, limits, psychology).
- Assuming Fibonacci always recovers in one win; many Fibonacci variants require multiple wins after a streak to net out.
- Using D’Alembert without a real edge; smaller steps reduce blowup risk but do not create profitability if entries are random.
- No explicit cycle definition (when does a sequence start/end?), causing you to mix regimes and contaminate results.
- Unit size chosen from impatience rather than bankroll math; the base unit is the main determinant of survivability.
- Ignoring costs (spread/fees/slippage), which hit hardest when stakes escalate.
- Switching progression mid-streak (common under stress), which often locks in the worst of both systems.
- Not matching to persona constraints:
- Conservative traders often fail by copying aggressive progressions and removing caps.
- Aggressive traders fail by overtrading cycles, not by choosing the “wrong” sequence.
- Algorithmic traders fail by omitting safeguards (cooldowns, max exposure) in code.
Execution guide: bet-sizing, stop rules and worked examples
For most intermediate traders in Thailand comparing these progressions (เปรียบเทียบ มาร์ติงเกล ฟีโบนัชชี ดาเลมแบร์), D’Alembert is often best for the conservative persona prioritizing stability and easy rule-following; capped Fibonacci is often best for the algorithmic persona needing bounded risk with structured recovery; a tightly capped, small-base Martingale can fit an aggressive persona who explicitly accepts cycle failures and manages exposure mechanically.
Practical concerns, pitfalls and myth-busting
Is Martingale ever “safe” if my win rate is high?
No; a high win rate reduces frequency of long losing streaks but does not remove them. Safety comes from a hard cap on steps and exposure, not from optimism about probability.
Does Fibonacci guarantee recovery with smaller risk than Martingale?
Fibonacci usually increases slower than doubling, but it can still reach large sizes in long streaks. It also may need more than one win to recover, depending on the step-back rule.
When is D’Alembert a bad fit?
If your edge relies on fast recovery after rare losses, D’Alembert may be too slow. It can also underperform when volatility regimes shift and the unit step is not adjusted.
Should I reset the sequence after any win?
Resetting reduces exposure quickly but can slow recovery if results alternate. Pick a reset/step-back rule you can follow consistently and test it on your specific entry/exit method.
How do I choose the base unit without overfitting?
Anchor it to a fixed bankroll fraction and then validate against worst-case streak assumptions up to your cap. If the worst-case cycle loss would change your behavior, the unit is too large.
Can I mix these systems (e.g., Fibonacci then switch to Martingale)?
Mixing mid-cycle is usually a discipline failure disguised as optimization. If you want a hybrid, define it upfront with fixed transition rules and a maximum exposure limit.
What is the minimum record-keeping that actually helps?
Log: date/time, instrument, step index, stake, outcome, and whether a cap/stop rule triggered. Without this, you cannot diagnose whether losses came from the edge, costs, or progression design.
