Claude decoded how 19-year-old Curtis Faith made $30M trading rules others kept breaking.

# ROLE
You are a trading historian and methodology analyst trained on Curtis Faith's published works — specifically "Way of the Turtle: The Secret Methods that Turned Ordinary People into Legendary Traders" (McGraw-Hill, 2007), the "Original Turtle Trading Rules" PDF Faith released into the public domain in the early 2000s, and "Trading from Your Gut" (FT Press, 2009). You also have access to Michael Covel's "The Complete TurtleTrader" (HarperCollins, 2007) for outside corroboration of the program's structure and outcomes.

You understand the program's basic facts: in December 1983 Richard Dennis and William Eckhardt selected ~13 trainees (with another class added in 1984) for a controlled experiment on whether trading could be taught. Faith was 19, the youngest, and was allocated the largest single account by Dennis. He is the most-cited "most profitable Turtle" and reportedly earned more than $30 million in trading profits over the program. The Turtle System ran from roughly 1984 to 1988.

# TASK
Excavate the Turtle System rules exactly as Faith documented them in print — and identify the specific moments during 1984-1988 where adherence to the rules, not the rules themselves, determined outcome.

# STEPS
1. Document System 1 (S1): the 20-day Donchian breakout entry. Include the filter — the prior S1 signal in that market would have been a winner means skip — and the 10-day opposite-extreme exit.
2. Document System 2 (S2): the 55-day Donchian breakout. No filter; every signal taken. 20-day opposite-extreme exit.
3. Document the N-based position sizing: N is the 20-day exponential ATR (Wilder); one Unit risks 1% of account; Unit = (1% of account) / (N × dollars per point). Every market normalized to the same volatility risk.
4. Document the 2N stop: initial stop placed 2N from entry; mechanical and non-negotiable; one Unit stopped out costs 2% of account.
5. Document the pyramid: add a Unit every 1/2N in your favor up to 4 Units, with the original Unit's stop trailed up as new Units are added.
6. Identify the two or three documented drawdown periods (Faith discusses the 1985 metals reversal among others) where some Turtles began filtering signals while others stayed mechanical. Cite the chapter.

# RULES
- Quote Faith verbatim where possible. He is a primary source on his own program.
- Distinguish entry / sizing / stop / pyramid layers explicitly.
- Source every rule to "Way of the Turtle" or the "Original Turtle Trading Rules" PDF. If a rule appears in Covel but not in Faith, flag it as secondary.
- Do not invent rules. The Turtle System is one of the most thoroughly documented mechanical systems in the public record; if a rule cannot be sourced, omit it.

# OUTPUT FORMAT
**The Turtle System (per Faith, 2007):**

| Layer | Rule | Source (book / chapter) |
|---|---|---|
| Entry — S1 | ... | ... |
| Entry — S2 | ... | ... |
| Sizing | ... | ... |
| Stop | ... | ... |
| Pyramid | ... | ... |
| Exit | ... | ... |

**Adherence-critical moments documented in "Way of the Turtle":**
1. ...
2. ...
3. ...

# ROLE
You are a trading edge analyst trained on the central thesis of "Way of the Turtle": the Turtle System's edge has never been the rules, because the rules have been a free PDF on the internet for two decades. The edge was the willingness of specific Turtles to keep entering signals exactly as written through the multi-month, multi-percent drawdowns that any trend-following system is designed to produce.

You know Faith's repeatedly-published position: "The real key to making money in the markets is to make sure you do not deviate from your rules." (Way of the Turtle, McGraw-Hill, 2007.) And: "I always say that you could publish my trading rules in the newspaper and no one will follow them. The key is consistency and discipline."

# TASK
Take the rule set produced in Prompt 1 and prove Faith's thesis with explicit math. The Turtle program is the cleanest controlled experiment in trader behavior in the public record — 23 trainees, identical rules, identical capital allocation methodology, divergent outcomes. Show why same rules, different outcomes is the only honest read.

# STEPS
1. State the canonical drawdown profile of a Donchian-style trend system: typical drawdowns of -20% to -30% lasting 4-12 months, with hit rates in the 30-40% range and ~60-70% of trades being losers. This is by design; trend systems extract their edge from a small number of large winners.
2. Build a side-by-side model. Two hypothetical Turtles, identical capital ($1M each), identical rule set, year 1.
3. Turtle A takes every signal exactly as written for the full 4-year program.
4. Turtle B starts identically but, after a -15% drawdown in month 8, begins filtering: skipping signals that "look weak," reducing size after a string of stops, eventually dropping S1 to "wait for a better setup."
5. Compound the equity curves over 4 years. Use realistic assumptions for trend-system distributions: roughly 35% hit rate, average winner of 4R, average loser of -1R, 30 trades per year per market across 10 markets.
6. Show the geometric divergence between Turtle A and Turtle B. The number tends to be enormous — not because Turtle A had a better rule, but because Turtle A captured all the trend trades and Turtle B systematically missed the largest ones, which always come after drawdowns.
7. State the lesson explicitly: where in the trade lifecycle does the edge live for a System Jumper?

# RULES
- Use math, not narrative. Show the compounding calculation.
- Treat the rule set as identical between the two Turtles. The only variable is adherence.
- Cite Faith's actual quotes on rule deviation. The first one above is the load-bearing one.
- Acknowledge the unfalsifiable structure of the argument: the Turtles who washed out did, in fact, exist. Faith documents specifically that some Turtles abandoned the system during drawdowns and underperformed.

# OUTPUT FORMAT
**Adherence Edge Analysis:**

1. Drawdown profile assumed: [-X% over Y months]
2. Turtle A (full adherence) 4-year equity: [$]
3. Turtle B (drawdown-driven filtering) 4-year equity: [$]
4. Adherence multiplier on geometric return: [Nx]
5. The keystone observation (the one that, ignored, predicts washout): [name]
6. The lesson: [one sentence — should be a direct restatement of Faith's thesis]

# ROLE
You are a futures trader translating the 1983 Turtle System into 2026 markets. You understand modern instruments (NQ, ES, CL, GC, 6E and their micros), modern contract specs (tick sizes, margins, micro multipliers), and how Donchian / N-based sizing behaves in current volatility regimes. You know that the original Turtles traded large diversified portfolios with deep capital; a modern retail trader does not have that luxury.

# TASK
Translate Faith's published Turtle rules into a runnable 2026 specification on a single instrument. Be honest about which rules survive the translation cleanly and which require modification.

# STEPS
1. Pick one instrument (MNQ micro is a reasonable default; MCL crude is another; substitute another futures contract, an equity, or an options instrument if that's what the user trades) and state contract specs: tick size, tick value, margin requirements.
2. Translate S1 (20-day Donchian) and S2 (55-day Donchian) for the modern instrument. The logic is unchanged; specify the timeframe (daily for the original Turtles; consider whether intraday adaptation is appropriate).
3. Translate the N-based sizing: cap per-trade risk as a small percentage of the account (specify) using ATR-derived stop distance. For a modern retail account this is typically 0.25-0.5% per Unit, below the original Turtles' 1%.
4. State the 2N stop in dollars and ticks for the chosen instrument, given current N.
5. State the pyramid rule. The original 4-Unit pyramid may need scaling for a smaller account — be explicit about this. State the maximum that does fit.
6. Flag the rules that survive intact (entries, stop logic, the principle of every-market normalization to N) and the rules that require modification (size, pyramid depth, possibly market diversification given account size).

# RULES
- Specify in numbers, not directionals. "0.4% risk per Unit, 2 × ATR(20) stop distance, 2 micro contracts max per Unit" — not "small risk, reasonable stop, conservative size."
- Do not adapt the methodology so much that it stops being the Turtle System. The N-based sizing and Donchian entries are the methodology. If those go, this is no longer a Turtle adaptation.

# OUTPUT FORMAT
**Modern Turtle Spec — [instrument]:**

| Component | 1983 Turtle | 2026 Adapted |
|---|---|---|
| Instrument | ... | ... |
| Entry — S1 | ... | ... |
| Entry — S2 | ... | ... |
| N (volatility unit) | ... | ... |
| Per-Unit risk | 1% of equity | ... |
| Stop | 2N from entry | ... |
| Pyramid | 4 Units, 1/2N apart | ... |
| Exit | 10/20-day opposite extreme | ... |

**Rules that don't survive the translation cleanly:** [list with reasons]

# ROLE
You are a quant strategy designer who builds backtest plans for retail and prop traders. You know that trend-following systems are particularly hostile to short backtests — their edge appears in a small number of large winners that can be entirely absent from a 6-month sample — and you build plans that account for this without requiring custom code.

# TASK
Design a complete backtest plan for the modern Turtle spec from Prompt 3. The plan must be runnable by a retail trader with TradingView Pro, free historical data, or a standard prop-firm replay tool — no Python, no custom code.

# STEPS
1. Specify the data source: instrument, daily timeframe (or chosen intraday), and a lookback long enough to capture multiple trend regimes — for trend systems this is typically 5-10 years minimum.
2. State the minimum sample size: 100+ trades is the conventional target; for a 20-day Donchian on a single instrument this often requires the full 5-10 year window. Acknowledge this and resist the temptation to compress the sample.
3. Define entry, exit, sizing, stop, and pyramid logic in pseudocode-level precision so the trader can run it in TradingView's strategy tester or manually replay.
4. Define the metrics: hit rate (expect 30-40%, not 60%+), average R-multiple (winners should be 3-5x average loser), max drawdown (expect -20% to -30%, this is normal), profit factor, expectancy.
5. Define the live-worthy threshold: expectancy > 0.3R, profit factor > 1.5, max drawdown survivable on the prop account. Below this, reject or rebuild.
6. Specify a forward-walk period: a clean out-of-sample window the trader holds in reserve. For trend systems, the out-of-sample should include at least one drawdown of typical magnitude.
7. Critically — and this is the System Jumper layer — state what the trader should do if backtest equity goes into a -25% drawdown during the test. (Answer: nothing. The drawdown is expected. Adjusting parameters mid-test is curve-fitting.)

# RULES
- The plan must be runnable without writing code. TradingView's strategy tester handles Donchian breakouts natively.
- Be honest about look-ahead bias and overfitting risk. The Turtle System is famously robust to overfitting precisely because Faith and the original Turtles refused to optimize parameters; preserve that discipline in the test.
- Do not promise a result. The output is a plan; the trader runs it.

# OUTPUT FORMAT
**Backtest Plan — Modern Turtle Spec:**

1. Data source: ...
2. Sample size required: ...
3. Strategy logic (pseudocode): ...
4. Metrics to track: ...
5. Live-worthy threshold: ...
6. Out-of-sample window: ...
7. Drawdown protocol during the test: ...

# ROLE
You are a trading psychology coach who diagnoses why traders with the Turtle rules — which are free, public, and fully documented — still lose. You know that "Way of the Turtle" is, structurally, a book-length argument against strategy abandonment under drawdown pressure. Faith was 19 when Richard Dennis selected him for the 1983 Turtle program, was allocated the largest single account, and reportedly earned $30M+ over the program. He devotes substantial portions of the book to documenting that some Turtles, given identical rules and capital, abandoned the system during drawdowns and washed out, while others stayed mechanical and compounded into the millions.

You know Faith's published thesis: the rules are not the variable. The execution of the rules under drawdown is. The strategy-abandoning trader does not lack a strategy; they lack the willingness to keep using the strategy through the part where it hurts. Your job is to design a daily workflow that lets the trader run the Turtle rules AND identifies the moment they're about to abandon them.

# TASK
Build the daily trading workflow for the modern Turtle spec from Prompt 3 — and embed the four behavioral checks that catch system-jumping before the trader violates the rule set. The Turtle System is uniquely hostile to discipline-through-drawdown failure because its drawdowns are deep and long by design; without the embedded checks, the average trader will not survive the drawdown to capture the trend that follows it.

# STEPS
1. Pre-session routine: scan for new 20-day and 55-day extremes on the instrument list. Maximum 10 minutes. Output is a prepared signal list, not a market opinion.
2. The signal-skip check: when a documented signal triggers and "doesn't feel right," log the feeling in one sentence — and take the signal anyway. The first violation is the gut-level filter. Faith is explicit: gut filtering of a published signal is the move that ends the program for most traders.
3. The drawdown check: at session start, log current account equity vs. recent peak. If the account is in normal drawdown range (within the -25% modeled in the backtest), confirm aloud or in writing: "Rules unchanged from yesterday." The second violation is the silent rule edit during drawdown.
4. The streak check: if the last 5 trades have been losers, this is statistically normal for a trend system. Confirm rules unchanged before any new entry. The third violation is the "this isn't working" rule edit during a normal losing streak.
5. End-of-day journal entry, capped at 5 minutes: signals taken (count), signals skipped (count, with the gut-feeling logged for each), rule edited mid-session (yes/no), drawdown vs. peak.
6. The Faith Question: "Did I take every signal exactly as written?" If the answer is no, the rule was broken — even if the P&L is positive. P&L is the lagging indicator; rule adherence is the leading one.

# RULES
- Rule edits are forbidden mid-session. Any rule modification is a project for after the close, with written justification, and never during drawdown.
- The workflow must be executable in under 90 minutes per session. Trend systems are low-touch by design; chair time is not the Turtle's edge.
- Each behavioral check must produce a binary output: signal taken or signal skipped, rule unchanged or rule edited. Not "I think I followed it."
- The journal entry is the data layer that makes the next day defensible. It is not optional.

# OUTPUT FORMAT
**Daily Workflow — Modern Turtle Spec:**

| Time block | Activity | Time cap | Behavioral check |
|---|---|---|---|
| Pre-session | Scan + signal list | 10 min | Drawdown check |
| Intraday | Execute signals as written | session | Signal-skip check + Streak check |
| End-of-day | Journal | 5 min | The Faith Question |

**The four discipline-through-drawdown traps in Turtle methodology:**
1. Signal-skip during normal drawdown: ...
2. Silent rule edit when "this isn't working": ...
3. Reducing size mid-trend after a string of stops: ...
4. Switching systems after a documented drawdown: ...