B.2 Building Your Own System

Codify your strategy into a written, testable system. Know when to follow rules, when to use discretion, and how to evolve the system over time.

Bonus Layer — Chapter 2 Goal: Codify your strategy into a written, testable system. Know when to follow rules, when to use discretion, and how to evolve the system over time.

The Core Idea

A trading "system" is a set of rules precise enough that you could hand them to another person and they would execute identical trades.

This is the test. If you can't articulate your strategy in writing — entries, exits, sizing, when to skip, when to engage — you don't have a system. You have a vibe. Vibes are unmeasurable, unreproducible, and unimprovable.

A written system is the foundation for every other professional skill: backtesting, journaling, statistical analysis, automation.

What a Written System Includes

The complete document has six sections:

1. Strategy thesis (1-2 paragraphs)

In plain English, what is the edge? Why does it work?

Example: "Stocks in clear uptrends (above rising 20EMA and 50EMA) that pull back to the 20EMA on declining volume tend to resume the uptrend more often than not. This works because institutional money uses the 20EMA as a key re-entry level after temporary weakness, and the volume contraction during pullback indicates lack of real selling pressure. The asymmetry between win rate and reward-to-risk creates positive expected value."

2. Universe definition

Which stocks do you trade? Rules for inclusion/exclusion.

Example:

U.S. equities only

Price > $20 (avoid penny stocks)

Avg daily volume > 1M shares OR > $50M dollar volume

Market cap > $2B

Listed for > 1 year (avoid newly IPO'd unknowns)

Excludes biotech (binary FDA risk is non-strategy)

Excludes Chinese ADRs (geopolitical and audit risk)

3. Entry criteria

Precise, no-ambiguity rules.

Example pullback setup:

Daily 20EMA > Daily 50EMA (uptrend)

20EMA rising (slope > 0 over last 5 days)

Price has pulled back ≤ 5% from recent swing high

Price touches or comes within 1% of 20EMA

Daily volume on pullback day < 20-day average

Bullish reversal candle (hammer, engulfing, etc.)

Next day: enter at open or on cross of prior day high

4. Risk and sizing

How much do you risk?

Example:

Fixed fractional at 1% of account per trade

Stop at 2× ATR below entry OR below 20EMA (whichever is wider)

Max simultaneous positions: 5

Max aggregate open risk: 4% of account

No more than 3 positions in same sector

Reduce risk to 0.5% per trade during > 10% drawdown

5. Management rules

What happens once you're in?

Example:

Move stop to breakeven once trade reaches +1.5R

Take 50% off at 1.5R if reached within 5 days

Trail remaining 50% below daily 20EMA

Time stop: exit if no progress within 10 trading days

Earnings: cut full position by close of day before earnings (re-enter after if setup re-establishes)

6. Exit rules

How and when do you fully exit?

Example:

Stop loss hit (full exit)

Trailing stop hit (full exit of remaining position)

Target 2 hit at 3R (sell remaining)

Time stop expired (full exit at market close)

Thesis invalidated (e.g., trend break) — full exit

Bonus section: When NOT to trade

Example:

VIX > 25 (high-vol regimes; pullback setups fail more often)

SPY below 200-day MA (bear regime)

Within 3 days of FOMC meeting

Personally tilted (after 2+ losses today or 5+ losses this week)

Less than 6 hours sleep

Setup is on a stock not in my universe

Why You Need It in Writing

Reason 1: Discipline under stress

Your strategy lives in your written document, not in your head. When you're stressed (loss, FOMO, fatigue), your brain will rationalize anything. The written document is the calm voice that wins the argument.

Reason 2: Measurement

Without precise rules, you can't measure your edge. "I trade good setups" can't be backtested. "I enter on a daily bullish engulfing at the 20EMA in an uptrending stock with rising 50EMA" can be backtested.

Reason 3: Improvement

You can only improve what you've defined. If your strategy is fuzzy, you can't tell whether a losing streak is a parameter problem, a discipline problem, or just variance. Defined strategies are debuggable.

Reason 4: Communication

If you want to learn from a mentor, partner with another trader, or eventually run capital for others — you need to articulate your system. The first version is for yourself; later versions are for others.

Rules vs Discretion

The eternal debate. Pure rules-based vs discretionary trading.

Pure rules (mechanical)

Backtestable
Automatable
Emotion-free execution
BUT: misses context the rules don't capture
AND: only as good as the ruleset (can be wrong/curve-fit)

Pure discretion

Adapts to context (market conditions, news, intuition)
Captures things rules can't easily encode
BUT: emotional execution
AND: not measurable; impossible to improve systematically

The pragmatic middle

Most successful retail traders use rules-based entries with discretionary management.

Entry rules are precise (this triggers, you take the trade)
Exit rules have defaults but allow some judgment (e.g., "trail below 20EMA unless major news breaks")
The default is to follow rules; deviation requires articulated reason

A useful framing

"The rules tell me what to do most of the time. My discretion is for the edge cases — but I should be skeptical of every edge case I find, and demand evidence that my discretion is better than the rule."

If 80%+ of your trades follow the rules exactly, you have a system. If 30% of your trades involve "judgment calls," you have a vibe.

How to Build Your First System

Step 1: Define the thesis (1 hour)

What do you think creates edge in the market? Write 2-3 paragraphs. Be specific about why this works.

Step 2: Define the universe (30 min)

Which stocks does this apply to? Write a filter rule. (Price, volume, market cap, sector.)

Step 3: Define entry (1 hour)

What does the chart look like at the moment of entry? Write the conditions precisely. Can a machine identify the setup? If not, sharpen the rules.

Step 4: Define risk (15 min)

Stop placement rule. Sizing rule. Max positions. Max aggregate risk.

Step 5: Define management (1 hour)

What happens after entry? Profit targets, stop moves, time stops, earnings handling.

Step 6: Define exits (15 min)

Conditions for full closeout.

Step 7: Define when not to trade (30 min)

Personal state, market conditions, account state.

Step 8: Save it (5 min)

Markdown file, Google Doc, Notion page. Whatever you'll actually read.

Step 9: Manually paper-test (4-8 hours)

Go through the last 6-12 months of charts on 20 stocks in your universe. Manually identify every trade signal. Calculate hypothetical results. Does it look profitable?

Step 10: Live with tiny size (50-100 trades)

Trade real money but at 1/4 normal size. Build the muscle of following the system without major drawdown if it doesn't work.

Step 11: Review and evolve

After 50-100 trades, statistical patterns emerge. Modify the system with data — not vibes.

When to Modify the System

Almost never on a short timeline. Most modifications you want to make are bias talking.

Good reasons to modify

50+ trade statistical patterns showing certain conditions fail
A new academic study or trader insight that maps to your strategy
Market regime shift confirmed over months (not days)
You've grown the account and risk percentages need scaling
Personal life circumstances change (more or less time available)

Bad reasons to modify

"I lost three in a row, the strategy doesn't work anymore"
"I read a Twitter thread saying RSI works better than MACD"
"My friend made money on biotech, I want to add it to my universe"
"I have a hunch this market is different"
"I want to try this new indicator I just learned"

The default is to NOT modify. Treat modifications like medical interventions — only when needed, with evidence.

Version Control

Track changes to your system over time. Date every revision.

Example versioning

v1.0 (Jan 2026): Original system. Pullback to 20EMA, 1% risk.

v1.1 (Feb 2026): Added "no trade if VIX > 25" filter after observing 5 losses during a vol spike.

v1.2 (Apr 2026): Increased risk to 1.25% after 100 trades showing 0.4R expectancy. Per Kelly check, this is still well below Quarter Kelly.

v1.3 (Jul 2026): Added breakout setup as second strategy. Universe and risk rules unchanged.

This history is invaluable. Two years from now, you'll want to know why you made each change.

Multiple Strategies

After your first strategy is profitable for ~100 trades, you might want to add a second.

Why add strategies

Reduces concentration on one setup
Different setups work in different market regimes
Diversifies "strategy risk" — when one is in a drawdown, another might be profitable

Rules for adding strategies

Don't add a second until the first is profitable for 100+ trades
The new strategy must have a different thesis (not just minor tweaks)
Track each strategy's stats separately
Sizing rules apply to TOTAL account, not per strategy (don't risk 1% on each = 2% total)
Limit to 2-3 strategies maximum until you're advanced

Common combinations

Pullback + breakout (both bullish, but different patterns)
Long swings + short swings (different regimes)
Trend trades + mean reversion (anti-correlated regimes)
Stocks + options for hedging (advanced)

When to Throw Out a System

Rarely. But sometimes.

Hard kill criteria (you wrote these in v1.0)

Drawdown exceeds X% (e.g., -25%)
Win rate over 100+ trades drops below threshold (e.g., 35% with 2:1 R:R = breakeven)
Profit factor drops below 1.0 over 100+ trades
Strategy stops behaving consistent with backtest (regime change)

Soft signals

You stop trusting the strategy emotionally
You're skipping trades that match the rules
You're adding trades that don't
Your journal shows you're "improvising"

If kill criteria hit, stop. Take a break. Examine what changed.

If soft signals hit, you might not have a strategy problem — you might have a discipline problem. Journal honestly and figure out which.

The "Boring" System Principle

The best strategies are usually boring.

They don't promise huge returns
They don't have 90% win rates
They don't use exotic indicators
They work in some conditions and not others
They require patience and discipline more than brilliance

If your strategy feels exciting, suspect it. Excitement comes from variance. Boring comes from edge.

Counter-intuitive truth

If your system regularly says "no trade today" on most days, that's a sign it's good. The boring days are when you're saving money by not forcing trades.

Sample Complete System Document

A simplified template:

# My Swing Trading System v1.0
Date: 2026-06-17
Author: [Your Name]

## Thesis
[2-3 paragraphs on edge and why it works]

## Universe
- US equities, price > $20, ADV > $50M, market cap > $2B
- Exclude: biotech, Chinese ADRs, recent IPOs (<1 yr)

## Entry: Pullback to 20EMA
1. Daily 20EMA > 50EMA
2. 20EMA rising (5-day slope > 0)
3. Pullback ≤ 5% from swing high
4. Touch within 1% of 20EMA
5. Pullback volume < 20-day average
6. Bullish reversal candle on day of EMA touch
7. Enter next day at open or above prior day high

## Risk
- 1% of account per trade
- Stop: 2× ATR below entry OR below 20EMA (wider)
- Max 5 simultaneous positions, max 4% aggregate risk
- Reduce to 0.5% in > 10% drawdown

## Management
- Move stop to BE at +1.5R
- Sell 50% at 1.5R if reached < 5 days
- Trail rest below daily 20EMA
- Time stop: 10 trading days

## Exit
- Stop hit
- Trail hit
- Time stop expired
- Trend break (close below 50EMA)
- Earnings: cut full position day before

## Don't Trade
- VIX > 25
- SPY below 200-day MA
- FOMC week
- After 2 losses today / 5 this week
- < 6 hours sleep
- Off-universe stocks

## Stats Targets (after 100 trades)
- Win rate ≥ 45%
- Avg win:loss ratio ≥ 1.5
- Profit factor ≥ 1.4
- Expectancy ≥ +0.3R per trade
- Max drawdown ≤ 15%

## Kill Criteria
- Drawdown > 25%
- Profit factor < 1.0 over 100 trades
- Strategy stops matching backtest behavior

## Change Log
- v1.0 (2026-06-17): Initial system

That's it. Two pages. Print it. Read it before every trade.

Common Mistakes

No written system. "I know what I do." You don't. Write it.
System too vague. "Buy strong stocks at good levels." Define strong. Define good.
System too complex. 20 indicators, 10 filters. Likely curve-fit; impossible to follow.
No "don't trade" conditions. You'll trade through every regime, including the wrong ones.
Constantly tweaking. Modifying after every losing streak. The system never gets enough sample size.
Ignoring the system in real time. You wrote it but you don't follow it.
No versioning. Changes lost to memory. Can't tell which version produced which results.
Stealing someone else's system. It won't fit your psychology, time, or account size. Build your own.
Skipping the paper-test step. Jumping to live trading with no manual validation.
No kill criteria. No way to know when to stop.

A Mental Model: The Recipe Card

A great chef doesn't improvise from scratch every time. They have recipes — precise instructions for dishes that work. Each recipe was developed and tested over many iterations.

When the chef is rushed, sick, or distracted, the recipe protects the dish. The recipe doesn't constrain creativity; it preserves quality.

Your trading system is your recipe. Precise enough to follow on bad days. Battle-tested over many iterations. Adjustable over time, but only with intention.

Practical Takeaways

Write your system before live trading at full size. This is non-negotiable.
Keep it under 2-3 pages. Complexity kills execution.
Define when NOT to trade as carefully as when to trade.
Paper-test or backtest at least 50 historical signals before live trading.
Live trade at 1/4 size for the first 50 trades before scaling up.
Version your system. Date every change.
Default to NOT modifying. Modifications need evidence.
Kill criteria are pre-defined. Don't decide in the moment.
One strategy until profitable for 100 trades. Then maybe add another.
Boring systems are good systems.

Quick Self-Check

I have (or will write today) a 1-3 page written system document
My system defines thesis, universe, entry, risk, management, exit, don't-trade
My rules are precise enough that another person could execute them
I have defined kill criteria for when to stop trading the system
I have a versioning approach for tracking changes
I will paper-test or backtest before going live
I will trade at 1/4 size for my first 50 trades
I default to NOT modifying the system without 50+ trades of evidence

Previous: B.1 Swing Trading Specifics Next: B.3 Automation