The previous lessons described setups and risk rules. None of them helps unless it is applied the same way every day and has been checked against history before money depends on it. This lesson turns the course into a working method: a routine that fits around a working day, a way to read backtest results, and the habits that keep a test honest.
The daily cycle
Swing trading on daily bars has a natural rhythm. Decisions are made after the close, when the day's bar is final, and executed in the next session. A complete cycle has four steps.
- Scan. Start with the regime: is the index above a rising 200-day average, how broad is the advance, is volatility rising? Check the economic and earnings calendar for the coming days. Then run your systems on the watchlist and list the setups armed for tomorrow and the entries triggered today.
- Plan. For each candidate, write the plan from lesson 1: setup, trigger, stop, exit, invalidation. Calculate the size from the stop, then run the event and portfolio checks from lesson 6. Discard anything that fails a check, however attractive the chart.
- Execute. Place orders that match the system's assumptions. A rule that enters on the close needs an order near the close; a rule tested with next-open fills needs an order at the open. Enter protective stops as soon as positions are filled.
- Review. Record every closed trade: setup, dates, entry, exit, reason for the exit, result in R, and whether every rule was followed. Once a week, look at the R results as a group and at rule breaks separately; a loss taken by the rules is not a mistake, while a profit taken against them is.
What a backtest tells you
A backtest replays a system's rules on past prices and records the trades they would have produced. It cannot show what will happen, but it can show whether the rules ever had an edge, what kind of profile they produce and how deep the bad periods were. Read these figures together (see also backtesting):
- Number of trades. Thirty trades say little; several hundred begin to separate skill from luck.
- Win rate, average win and average loss. Each is meaningless alone. Together they describe the profile.
- Expectancy. The average result per trade: win rate × average win − loss rate × average loss. Expressed in R, it compares systems with different stop distances.
- Profit factor. Gross profits divided by gross losses. Above 1 the system made money; values well above 2 in a backtest deserve suspicion.
- Maximum drawdown. The largest fall from a peak in equity. Expect a live drawdown at least as deep.
- Time in the market and the benchmark. A system that earns less than buying and holding the index, while being invested only part of the time, may still be useful; one that earns less while fully invested probably is not.
Two hypothetical systems show why the profile matters:
| Trend system | Mean-reversion system | |
|---|---|---|
| Win rate | 40% | 70% |
| Average win | 2.5R | 0.8R |
| Average loss | 1.0R | 1.2R |
| Expectancy | 0.4 × 2.5 − 0.6 × 1.0 = +0.40R | 0.7 × 0.8 − 0.3 × 1.2 = +0.20R |
| Profit factor | 1.00 ÷ 0.60 = 1.67 | 0.56 ÷ 0.36 = 1.56 |
Both are profitable on paper. The trend system will feel worse, with long strings of losses; the mean-reversion system will feel better until a cluster of large losses arrives. Knowing which profile you are trading prevents abandoning a sound system during its normal bad stretch.
How backtests flatter a system
- Look-ahead bias. A rule that decides on the close cannot always trade at that exact close. Testing with next-open fills is the conservative check.
- Survivorship bias. Testing only stocks that are listed today leaves out companies that were delisted or went bankrupt, often exactly the ones a dip-buying system would have bought.
- Costs and slippage. Short-term systems with small average gains can lose most of their edge to costs. Test with realistic costs, then with higher ones.
- A single favorable period. Almost any long-only system looked good in a steady bull market. Test across periods that include a bear market and a sideways year.
- Overfitting. The more variations you try, the more likely the best one won by chance.
Avoiding overfitting
Overfitting means tuning rules to the noise of the past rather than to a lasting pattern. The defense is not to stop testing but to test in ways that expose it.
- Keep rules few and explainable. Each rule should have a reason that exists outside the test, such as a trend filter that keeps you out of falling stocks.
- Look for plateaus. If 9, 10, 11 and 12 all work similarly, the idea is robust. If only 11 works, it is probably noise.
- Hold data back. Develop the rules on one period and test them once on a later period you have not looked at. A walk-forward test repeats this with a window that moves through time.
- Test many stocks. A rule that works on one ticker is an anecdote; one that works across hundreds is evidence.
- Count your attempts. If you tried forty versions, expect the best to overstate the truth.
From backtest to live trading
Expect live results to be somewhat worse than the backtest: fills are less clean, costs creep in and the market changes. Start with smaller size, keep the journal from the first trade, and compare the live distribution of R results with the tested one after a meaningful number of trades. Decide in advance what would make you stop, for example a drawdown 1.5 times the deepest one in the test, so the decision is not made in the middle of a losing streak.
How to apply it
- Every evening: regime check, calendar check, run the systems, write a plan for each candidate, size it, run the portfolio check, prepare orders.
- Every trade: enter the protective stop at once; exit only by the rules; record the R multiple and any rule break.
- Every week: review the R results and the rule breaks; check whether the regime still suits the systems you run.
- Before trading a system: test it over several periods, with conservative fills and realistic costs, across many stocks, and confirm it on data you did not use to build it.
In Clear Tape
Trading systems supports the whole cycle. You can switch on up to five systems, presets or your own, and keep a watchlist of up to 50 stocks; after each close, Tools and Pro members see the watchlist screened, with each signal marked as triggered or as a setup for the next session, together with its entry, stop and target. The system builder lets any member write rules from prices, moving averages, RSI, ATR and volume, with up to six entry and six exit conditions.
Backtests run any preset or your own system on daily prices over the last 1, 3, 5 or 10 years, on your watchlist, a single ticker or the screener's 500 stocks. You choose the entry fill, either the next day's open, labeled conservative, which skips a trade if that open is already at or below the stop, or the signal day's close; costs of 0 to 100 basis points per side, slippage included; and risk per trade of 0.1% to 5% of equity. The test starts with $100,000, caps each position at 20% of equity and uses no margin. Results show CAGR, total return, maximum drawdown, win rate, profit factor, average R before costs, expectancy after costs, average days held and time in the market, with an equity curve against SPY buy and hold and the trade list with each exit reason: stop, target, exit rule or time limit. Prices are split-adjusted, and dividends are left out for the system and SPY alike. Each report warns that only stocks listed today are tested, and a test on the screener's list adds that it consists of today's most liquid stocks, chosen with hindsight.
A sound sequence for any preset in this course: test it on the screener's 500 stocks over 10 years with next-open fills; repeat with higher costs; compare the 3-year and 10-year results to see how much depends on the period; then compare it with a variant you built yourself, changing one parameter at a time and looking for a plateau. Backtests, 20 a day, and screening are available to Tools and Pro members.
Common mistakes
- Trading a system you have not tested, or one you tested only on the stocks you already like.
- Testing with the most flattering settings: signal-day fills, no costs, one bull market.
- Changing the rules after every loss. Each change resets what you know about the system.
- Keeping no journal. Without recorded R multiples and rule breaks, you cannot tell a bad system from bad execution.
- Running too many systems at once. A small number you understand well beats a large number you cannot monitor.
Summary
A swing method is a routine: scan after the close, plan each trade in writing, execute what was planned and review every result in R. Backtests show whether rules ever had an edge and what profile they produce, provided they use conservative fills, realistic costs, many stocks and data held back from development. This completes the course: two families of swing trades, five setups mapped to Clear Tape presets, the risk rules they share and a way to test them. For the related tools, see relative strength, market breadth and risk/reward ratio.
For educational purposes only. Not investment advice.