Backtesting and Paper Trading Explained

Backtesting shows how a set of trading rules would have behaved on past prices. Paper trading lets you place fake-money trades as the market moves. Both are useful ways to learn a platform or test a paid tool before you commit, and both can mislead if you treat the results as proof. This guide explains how they work on TradingView, thinkorswim and broker paper accounts, where they fall short, and how to use them before you pay for a tool.

Nothing here is financial advice. Simulated results don't predict real ones.

Backtesting, forward testing and paper trading#

• Backtesting runs a strategy's rules over historical data. On TradingView, a strategy script does this and shows a Strategy Report with metrics and a list of every trade.
• Forward testing keeps the same strategy running on live, updating data, so you see how it behaves on bars it has never seen.
• Paper trading is you placing simulated orders with virtual money, by hand or from a tool's alerts, in real time.
• Replay sits in between: you step through past prices bar by bar and make decisions as if it were live.

TradingView: Strategy Tester and Bar Replay#

Strategy Tester#

Add a strategy to a chart and TradingView backtests it automatically. The results depend heavily on the strategy properties: initial capital (1,000,000 by default), order size, commission, slippage, margin and how orders fill. A Bar Magnifier option uses lower-timeframe prices to make historical fills more realistic.

TradingView's own publishing rules show what realistic looks like. They ask for a realistic starting balance, commission and slippage that match the instrument's real costs, and, in most cases, more than 100 trades in the report. They also call zero margin unrealistic.

Bar Replay#

Bar Replay takes a chart back to a past bar you choose, then plays it forward or steps one bar at a time. It's useful for practicing reading a chart, or watching an indicator draw bar by bar, without seeing what comes next. Two catches: paper trades you place during replay fill at today's real-time prices, not the replayed ones, and you can't create new server-side alerts while replaying.

How far back you can replay depends on your plan, as listed in October 2026:

Longer intervals reach further back on Essential and Plus. Plan prices are in our guide to trading indicators and algos.

Paper Trading#

TradingView says its Paper Trading is available to all users with no fees, across most of the markets it charts.

thinkorswim: paperMoney, OnDemand and thinkBack#

• paperMoney is Schwab's simulator. Schwab says it needs a Schwab account, is free for clients, starts with $100,000 of virtual buying power and uses real-time data for stocks, options, futures and forex.
• thinkOnDemand (OnDemand in the platform) replays trading periods from the last several months, so you can place simulated trades on past data.
• thinkBack is an options back-testing tool with nearly a decade of historical option data. Its results are hypothetical.

Other brokers' paper accounts#

• Interactive Brokers: paper accounts start with $1,000,000 of paper equity and use your live account's data subscriptions. IBKR says fills are simulated from the top of the book, some order types aren't supported, and stops and complex orders are always simulated, so behavior can differ from live trading.
• Webull: paper trading runs on its app, desktop and web platforms for stocks, ETFs, options, crypto and futures, and can be reset. On its legacy paper system, data is cleared after 180 days of inactivity.

Why a good-looking backtest can mislead#

Overfitting#

Tweak a strategy's settings enough and it will fit almost any stretch of history, then fail on new data. Researchers argue that the usual safeguards, including holding back data to test on later, are unreliable for investment backtests. The more settings a strategy has, and the more versions were tried, the less a single good backtest means.

Look-ahead bias#

A backtest that uses information that wasn't available at the time will look far better than it could ever trade. TradingView's docs give a common example: requesting higher-timeframe data in a way that pulls future values into historical bars, which they call dangerously misleading. Repainting strategies have the same problem: their backtests don't represent how they behave live.

Survivorship bias#

If you test a stock strategy only on companies that exist today, you leave out the ones that were delisted, merged or went bankrupt during the test period. The results look better than an investor trading at the time would have seen.

Costs and fills#

Backtests and paper accounts both assume fills. A limit order "filled" as soon as price touched it, or a market order filled with no slippage, flatters results, especially on fast intraday trades and options with wide spreads. Zero commissions do the same. Paper accounts simulate fills too, and Schwab says simulated performance doesn't ensure success in a live environment.

Why a seller's backtest isn't proof#

Indicator sellers often market with annotated chart screenshots and examples of past trades. Some share Strategy Tester reports or backtest summaries, more often for full strategies than for indicators that only plot alerts. A screenshot shows the trades the seller chose to show, so on its own it proves very little.

A credible test tells you:

• the market and timeframe tested;
• the date range;
• the exact entry and exit rules;
• the commission and slippage assumed;
• the number of trades and the maximum drawdown;
• whether the results are backtested, paper or live.

Then check it for the problems above: repainting, look-ahead bias, overfitting and cherry-picked examples. Treat any results as marketing, for three reasons:

• Regulated firms face rules on this. The SEC's marketing rule lets investment advisers show hypothetical performance only under conditions, including keeping it relevant to the audience and explaining how it was produced. FINRA's rule for broker-dealers bars communications that predict or project performance or imply past performance will recur. A tool seller that isn't a registered firm isn't covered by those rules.
• TradingView bars unrealistic or unsubstantiated performance claims about published scripts, on TradingView or elsewhere. A rule isn't verification, though: nobody checks a seller's backtest before you see it.
• Simulated accounts can be faked. A CFTC advisory describes a fraud whose founder created fake customer accounts and balances using demo accounts.

Our guide to spotting trading scams covers the bigger red flags.

How to test a tool on paper before you pay#

1. Use the trial, or a monthly plan, not a lifetime deal, so you can walk away.
2. Set up the alerts on a paper account and run them for a few weeks, ideally across different market conditions.
3. Log every alert: the time, the price, and what you'd have done. Compare the live alerts with what the tool's history shows for the same period. Gaps suggest repainting.
4. Add real costs: commissions, the spread and realistic slippage, plus the subscription itself.
5. Check plan limits. Alert counts and replay depth depend on your charting plan.
6. Practice the instrument first. If the tool is built for options or flow, our guides to options basics and options flow and scanners explain what you're looking at.

Testing tools on OOSClub#

OOSClub lists trading software on the tools hub and in the stocks and options niche, with each tool's platforms, plans, trials and refund terms. We never publish a tool's backtests or results; any results claim on a listing is labeled as the seller's.