Every strategy developer ends up in the same place: four or five variants of the same idea — different stop distances, a filter added or removed, a longer lookback — and a decision to make between them. The workflow most people use is flipping between Strategy Tester runs, eyeballing net profit, and keeping the mental leaderboard in their head. That workflow is error-prone in specific, predictable ways. This guide covers why, and lays out a repeatable process for comparing multiple TradingView strategies honestly on a unified time axis.
Why tab-flipping comparison fails
The tester shows one strategy at a time
TradingView's Strategy Tester renders the results of the strategy currently on the chart. Comparing five variants means five separate runs viewed sequentially, with your memory as the comparison layer. Human memory is a terrible comparison layer: the variant you looked at most recently reads as more vivid, and a big green number seen ten minutes ago gets rounded to "pretty good."
Different test windows masquerade as different performance
This is the subtle one. Two variants tested at different times, on different timeframes, or after a chart-history change may cover different date ranges — and the difference is easy to miss because the report doesn't exactly shout about it. Variant A tested over 2022–2026 and variant B tested over 2024–2026 are not comparable on any metric, yet their reports sit side by side looking equally authoritative. If B's window happens to exclude a brutal 2022 drawdown that A had to survive, B looks better for reasons that have nothing to do with the strategy.
Single-number comparison rewards the wrong winner
Even with identical windows, ranking variants by net profit alone systematically favors the variant that took the most risk or got the luckiest outlier. A comparison that doesn't include drawdown, risk-adjusted ratios, and some measure of consistency is a coin-sort with extra steps. (What each of those metrics actually describes — and their failure modes — is covered in TradingView Backtest Metrics Explained.)
The repeatable workflow
Step 1: Export every variant to .xlsx
Run each variant and export the Strategy Tester report. Name the files descriptively at save time — ORB-v1_baseline.xlsx, ORB-v2_atr-stop.xlsx, ORB-v3_volume-filter.xlsx — because "final_final2.xlsx" archaeology is real. The export mechanics and the parsing traps (paired entry/exit rows, open positions written as fake exits, no timezone metadata) are covered in How to Export TradingView Strategy Tester Results to Excel. Two comparison-specific additions:
- Export all variants from the same chart timezone. Timestamps carry no timezone metadata, so mixed-timezone exports silently misalign any time-based comparison.
- Export them in one sitting. Same day, same chart setup — this keeps the test windows as close to identical as your plan's bar history allows.
Step 2: Put every variant on a unified time axis
This is the single highest-value move in the whole workflow. Align all equity curves and period buckets to one shared date range — the intersection of the variants' test windows — so every strategy is judged on the same days, the same regimes, the same shocks. Once curves share an axis, questions that were invisible in tab-flipping become obvious at a glance: Did all variants draw down together in the same month (they share a regime sensitivity), or did one sail through what sank the others? Did the "winner" build its lead steadily, or in one spike? DataViz Studio's multi-strategy comparison (a Pro feature) renders uploaded datasets side by side on exactly this kind of unified time axis; in a spreadsheet, the equivalent is reindexing each variant's daily P&L to one shared date column before charting.
Step 3: Compare on a small, fixed metric panel
Decide the metric set before you look at results, and apply it identically to every variant:
| Metric | What it catches in a comparison |
|---|---|
| Trade count | Variants with too few trades to describe anything |
| Profit factor | Efficiency of the edge, independent of position count |
| Max drawdown | The historical worst stretch each variant produced |
| Sharpe / Sortino | Return per unit of risk; the gap between them shows volatility shape |
| Win rate × avg win/loss | Whether variants win the same way or via different trade-offs |
A useful habit: when one variant leads on net profit but trails on Sortino and drawdown, say out loud what that means — it made more by riding through more pain. Sometimes that's a trade-off worth understanding; it should never be an accident.
Step 4: Compare time-of-day signatures
Variants of one strategy often differ most in when they make money. An added filter might specifically kill midday trades; a wider stop might change which hours survive. Bucketing each variant's trades by entry hour — and checking whether each hour's result is persistent across rolling windows rather than a one-lucky-month artifact — adds a comparison dimension that aggregate panels miss entirely. The full hour-bucketing and Stability % method is in How to Find Your Strategy's Most Profitable Trading Hours.
Step 5: Check consistency, not just totals
Two variants with identical net profit can have opposite consistency profiles — one earned steadily across the sample, the other in a single hot quarter. Rolling-window analysis (what share of windows was each variant net profitable in?) separates them. In a comparison, the persistent variant and the concentrated variant are different objects even when their headline numbers match; a comparison that can't see the difference will keep picking lottery tickets over edges.
Step 6: Read the drawdowns as events, not numbers
With curves on a shared axis, drawdowns stop being abstract percentages and become dated events you can interrogate. When variants drew down simultaneously, the cause usually lives in the market — a regime shift, a volatility spike — and no amount of parameter tweaking was going to dodge it. When one variant drew down alone, the cause lives in that variant's specific logic, which is exactly the kind of difference a comparison exists to surface. This event-level reading is impossible when each report shows its drawdown as a single context-free number.
Keeping the comparison honest
- Compare on the intersection window. If one variant has less history, truncate everything to the shared range. An unshared month is an unfair month.
- Watch for one-trade winners. Recompute each variant's profit factor with its single largest winner removed. Rankings that reshuffle under this test were rankings of luck.
- Beware comparison-as-optimization. Picking the best of twenty variants on the same sample is implicit curve-fitting; the winner's edge is partly selection bias. The more variants you compare, the more weight consistency metrics deserve over totals.
- Record the decision. Note which variant you picked and why. When you re-export with fresh data later, you can check whether the reasons held up.
Tooling
Everything above is doable in a spreadsheet with enough patience — the unified axis is a reindex, the metric panel is a set of formulas repeated per file. If you'd rather skip the plumbing: each variant can be checked individually in seconds via the free instant report (no signup, processed in memory, never stored), and a Pro account ($19/mo, 7-day free trial) handles the full workflow — up to 20 datasets, side-by-side comparison on a unified time axis, Sharpe/Sortino, the Rolling Window stability heatmap per variant, and Excel export of the results. A free account supports 2 datasets with core charts and metrics, which is enough to run a serious two-variant comparison end to end.
The principle underneath all of it: a comparison is only as good as what it holds constant. Same window, same metrics, same axis — then, and only then, do the differences you see belong to the strategies rather than to the measurement.