TestingView logo TestingView.com

Strategy testing

What backtesting hundreds of trading strategies actually shows

21 October 2026 · TestingView · 12 min read

Backtesting hundreds of technical trading strategies teaches one hard pattern. A short sample can make almost any rule look profitable. Stretch those same rules across years, include spreads, commissions, and slippage, and most equity curves break. Use something like 20–30% a year after costs, with a drawdown you can sit through, as an illustration of a result worth taking seriously, not as a benchmark or a promise. A holy grail that prints 100% every year does not survive that test.

Treat that range as a filter your own journal still has to confirm. Markets, fees, and position size change the number. Use the test to decide what you fund.

Key takeaways

  • A few strong months, or even one strong year, are a trailer. The full backtest is the multi-year path after costs.
  • A sample of 100 trades, or 1,000 trades inside one quiet regime, can still be luck.
  • Chasing a 100% year often means accepting the drawdown that later gives the account back.
  • A repeatable 20–30% a year after commissions, spreads, and slippage is an illustration of a strong, survivable result, not a typical one or a promise.
  • Drawdown, profit factor, and expectancy matter as much as the headline return.
  • Rules you tweak until the past looks perfect are overfit. Freeze them, then test a fresh window.
  • The practical tool is bar-by-bar replay on your own TradingView chart, with a journal that records every loser.

Why a strong backtest collapses when you zoom out

Backtesting means taking a written strategy, running it on historical prices, and recording what those rules would have done. Open a chart, apply the rules, and the first window often looks brilliant. Two months of clean trends will flatter a breakout rule. A grinding year will flatter a mean-reversion rule. Volatility then changes, and the same entries start paying the market back.

Traders who work through a large stack of technical ideas see the same movie. The curve is wonderful over a few months. It is still wonderful over a full year. Add the next year, and the strategy is losing more than it wins. A rule that makes 250% in one hot year can wipe the account in the next. The thumbnail keeps the first number. Your capital lives through the second.

Markets are ruthless across regimes. Trend, chop, high volatility, and dead volume are different games. A strategy with no rule for “stand aside” will trade all of them. The long backtest is the only place that bill comes due before it comes due in the account.

A YouTube sample of 100 trades is a trailer

Public strategy videos love a round number. Tested 100 times. Tested 1,000 times. The win rate on screen is high, the narrator is calm, and the window is the window that behaved. That sample can sit entirely inside one mood of the market. It can also be the best slice of a longer history that was left off the chart.

One hundred trades is a start. It is enough to see whether the rule is even tradable. It is a thin basis for an annual return. One thousand trades taken while volatility stayed friendly still describe one regime. The next regime is a new sample, and it does not owe you the old win rate.

There is a second trap. If you invent a large pile of random rule combinations and backtest all of them, some will make money by chance. The winner of that search is often noise that happened to fit the file. Publish that curve, and it looks like skill. Trade it forward, and the edge was the search itself.

Demand the boring details before you fund a strategy you saw online:

  • The full date range, including the ugly years
  • Costs: commission, spread, and slippage
  • Maximum drawdown, and how long it lasted
  • Profit factor and expectancy, along with win rate
  • Whether the rules were frozen before the final test window

The annual return that actually survives

Take roughly 20–30% a year after commissions, spreads, and other fees as an illustration, not a benchmark: a result in that range that survives the bad years is real, and its risk is still something a trader can carry. Push the target to 100% or more and the strategy usually buys that number with a drawdown that later takes the account.

Traders hear 20–30% and call it small. Compounding changes that: an account that compounds near 25% doubles in a little over three years, if the path holds. Professional records that beat broad markets by a steady margin for many years are rare. A personal strategy that truly holds that band, after costs, is a serious result.

The percentage is also a function of risk. The same entry rules produce a different annual return if you risk 0.5% of the account or 5%. A higher risk percent raises the return and raises the chance of ruin. Judge the return next to the drawdown that produced it. A smooth 22% with a 12% drawdown is a different product from a 22% that sat through a 45% hole.

Technical analysis is the domain of this lesson. Other approaches, including deeper fundamental work, can print different distributions. For chart rules, indicator rules, and price-action checklists, the long test keeps pointing at sustainability. Consistency beats the flashy year.

Spreads, commissions, and slippage are part of the strategy

A fill on a backtest is a fantasy until the cost of trading is in the number. Commission is the obvious fee. The spread is the gap between the price you wanted and the price you could have. Slippage is the extra distance on a fast candle, a news bar, or a thin book. High-frequency rules die here first. A strategy that looks profitable at 200 trades a year can be a loser at the same signals once each trade pays the spread.

The built-in TradingView strategy tester is excellent at running a coded script quickly across history. It reports the profit you asked it to report. Leave commission at zero, leave slippage at zero, and every scalp looks richer than the live ticket. Traders who replay by hand on the chart have the same leak if they log the ideal price and ignore the spread they actually pay.

Put the costs in the test before you compare strategies. A 20–30% path that remains standing after those costs is the path worth your attention. A 90% win rate that disappears when each trade pays a realistic spread was never a strategy. It was a pricing error.

TestingView is the TradingView backtesting extension built for that measurement. On the Trader plan, slippage and spread go into the statistics, so profit, expectancy, and the equity curve already carry the cost of getting filled. You stay on the chart. You keep the indicators you trade with. The journal stops being a spreadsheet you dread opening.

Drawdown decides whether you can live with the return

Return without drawdown is advertising. The question under every backtest is simple: could you have followed the rules through the worst stretch in the file? If the answer is no, the annual percentage is a number you will never collect. You will pause, override, or quit at the bottom, which is a different strategy from the one you tested.

Read these stats together:

  • Win rate. How often a trade closes green. A 40% win rate can be excellent with a strong reward multiple. A 70% win rate can still lose if the losers are huge.
  • Profit factor. Gross profit divided by gross loss. Above 1.0 the file made money. A thin margin above 1.0 often vanishes once costs are honest.
  • Expectancy. The average result per trade. This is the number that scales. A tiny edge times too many trades can be eaten by spread.
  • Drawdown and run-up. How far the trade, and the account, moved against you and for you. The account drawdown is the bill for the annual return.
  • Streaks. Losing runs happen inside profitable systems. If six losers in a row would break your rules or your nerve, the system is too large for the risk percent you chose.

Position size is the lever. Risk a small, repeatable slice of the session until the record is long. Kelly-style math can justify a larger fraction when an edge is truly known. Most traders do not have an edge known that precisely. A fraction of a conservative risk keeps a 20–30% path alive. Full-size risk on a curve you have only seen once is how a good year becomes a wipeout.

Time of day belongs in the same review. Many strategies only pay in a specific session. Results by hour, by weekday, and by month show the hours that fund the return and the hours that donate it back. Cutting the dead hours is often worth more than a new indicator.

Overfitting is how a dead rule learns to look alive

Overfitting means shaping the rules until the historical file looks profitable. Add a filter. Nudge the RSI level. Swap the moving-average length. Refuse the trades that would have hurt. The curve smooths out. You have described the past with extra digits. You have a fresh sample still waiting, and that sample did not help you pick the parameters.

Lagging tools such as RSI and moving averages are a common raw material for this. They summarize price that has already printed. Mix them, optimize the periods, and you can avoid every big historical loss on purpose. The fit is real on that chart. Predictive power on the next chart is a separate claim, and it has to be earned on dates you locked away before the tweaking started.

A mechanical test also describes a mechanical trader. If your live edge is an early exit you “just see,” the replay will miss it. Write that discretion down as a rule you can apply with the next candle hidden, or accept that the backtest is about a stricter trader than the one who shows up live. The gap between the journal and the later forward test is the report on that difference.

Market context belongs in the rules too. A signal in a trend and the same signal in a chop are different trades. If you only take the setup at certain levels, certain hours, or certain conditions, those filters have to be in the test from the start. Adding them after you see the losers is another form of fitting.

Keep more than one playbook if you want. A strategy can work, sleep through a regime, and work again. Know which condition each playbook is for. On a funded path, a sleeping strategy is cheaper than a strategy you force through a market it was built to avoid. Test each one on its own session so the stats stay clean.

How to backtest a trading strategy on TradingView

The useful test happens on the chart you already trade. Your indicators, your levels, and your layout stay put. TestingView replays historical candles forward on that screen, for Chrome, Edge, Firefox, and TradingView Desktop. Future prices stay hidden. You take the trade with the same risk rules you intend to use live. The journal writes the row so a giant spreadsheet never becomes the reason you quit.

Run the session in this order:

  1. Open your TradingView layout and start a TestingView session. Set the account size, the currency, and the date the replay begins.
  2. Hide the future. Leave upcoming candles covered, and turn on Prevent Candle Previews so a symbol or timeframe change cannot spoil the move.
  3. Refuse the easy window. ChronoShuffle starts the replay at a random date inside a range you choose, which stops you from beginning on the month you already know.
  4. Trade only the written setup. Size the position from a percent of the session or from a fixed money risk, using the contract size for that symbol.
  5. Log every valid trade, including the ugly ones. The journal keeps price, symbol, lots, risk-reward, profit, cumulative profit, run-up, and drawdown.
  6. Include the cost of the fill. On the Trader plan, slippage and spread sit in the stats, so the curve is closer to a ticket than to a midpoint price.
  7. Read win rate, profit factor, expectancy, streaks, and the equity curve. Then read results by hour, day, and month. Optimal stop and target studies are there when you are ready to question the distances.
  8. Repeat on a new slice of history. When the rules still hold, forward-test them on unseen price before you raise size. A weekly replay is how that record stays honest. The habit is covered in regular backtesting on TradingView.

Sessions save and resume, including across devices on the same account. Snapshots can pin the TradingView chart to the trade, so a later review shows what you saw at the entry. Notes sit on the row. When you want the file outside the journal, the Trader plan exports it to Excel. You open the sheet to study. You do not build the sheet in order to remember.

Starter includes core backtesting and session resume, with a limit on trades in the session. Explorer raises that limit and adds break-even, advanced stats, and journal customization. Trader adds multiple sessions, Excel export, slippage and spread, and lot edits after entry. Pick the plan that matches how many strategies you intend to put through a real sample. See the plans, or start from the knowledge base if you want the controls first.

Why 20–30% is enough when the account is funded

A personal account and a prop-firm account do different jobs with the same percentage. On a small balance, 25% is meaningful and still slow as a salary. On a funded balance, that same 25% is a living, provided the drawdown stays inside the firm’s loss limit and you actually get paid.

This is why the boring backtest is the one that maps to a challenge. Model the daily loss limit. Model the overall drawdown. Model the fees and the spread. A strategy that needs a 40% hole to deliver its annual return will fail the firm even if the backtest “made money.” A strategy that compounds in the 20–30% band with a shallow drawdown is the one a funded account can carry.

A challenge fee is a cost, the same way commission is a cost. Paying for evaluations on an untested rule is a second strategy: buying tickets. Test the path first, on the chart, with the risk percent you will be allowed to use. Then the evaluation is a forward test of a record you already trust.

There is still no holy grail in the file. There is a sustainable strategy, a size you can repeat, and a journal that will tell you when the regime has changed. That is the whole edge the long test is willing to give you. TestingView is how you collect it without leaving TradingView.

What to require before you fund the next strategy

Use this checklist on every idea, including your favorite one:

  • The rules are written, and every trade in the test obeyed them.
  • The date range includes at least one ugly regime, and a second window you did not use to build the rules.
  • Commission, spread, and slippage are in the result.
  • You know the maximum drawdown, the longest losing streak, and the expectancy per trade.
  • The annual path you are willing to trade comes from your own long, cost-inclusive test. Treat 20–30% as an illustration, not a target.
  • Position size keeps a historical losing streak inside a loss you would still trade through, and inside any prop-firm limit you care about.
  • A forward test on unseen price is on the calendar before size goes up.

If a rule fails the list, it can still teach you something. It is an idea. Fund the ideas that survive the list. The fastest way to build that list is a replay session on the chart you already have open.

Questions traders ask about backtesting results

What does backtesting hundreds of trading strategies show?

A long test of technical strategies usually shows that a short winning streak fades when the window gets longer. After spreads, commissions, and slippage, far fewer paths keep working. Something like 20 to 30 percent a year with a survivable drawdown is used here as an illustration of a strong result, not as a typical outcome. A year of 100 percent or more often arrives with a later wipeout.

Is a 20 to 30 percent annual trading return a good result?

A 20 to 30 percent annual return is a strong result when it repeats after costs and the drawdown is survivable. It can compound into a large account, and on a funded account that same percentage is applied to a much larger balance. A single 100 percent year is a weaker signal.

Why do YouTube trading strategies fail a longer backtest?

A video sample is often a few dozen or a few hundred trades inside one market mood. Extend the dates, change volatility, and include fees, and the equity curve frequently rolls over. A small sample can crown a lucky rule.

Does a TradingView backtest include spreads and commissions?

A TradingView backtest includes spreads and commissions only when you put them in the test. A run with zero commission and no slippage overstates profit. TestingView on the Trader plan can apply slippage and spread inside the statistics so the result is closer to a real fill.

How do you backtest a strategy on TradingView without seeing future candles?

Replay history one step at a time with future candles hidden. TestingView plays the chart forward on TradingView, can open the session at a random past date with ChronoShuffle, and writes each trade to a journal with win rate, profit factor, drawdown, and expectancy.

How many trades do you need before you trust a backtest?

You need enough trades to see more than one market regime, including enough losers to measure drawdown. One hundred trades is a start. A strategy tested inside a single quiet year is still a hypothesis until you forward-test the same rules on unseen price.

What is overfitting in a trading backtest?

Overfitting means adding or tweaking rules until past prices look profitable. The equity curve then describes that history. Freeze the rules and test a fresh date range you did not use while building them.

Can backtesting help with a prop firm challenge?

Backtesting can show whether your rules, position size, and costs stay inside a drawdown a funded account would allow. A challenge pass is never guaranteed. Model the loss limit and the fees before you pay for an evaluation.