Find ideas. Test strategies. Understand the evidence.
Put your investing ideas to the test.
Discover what insiders and Congress are buying, research a question against real historical data, and backtest a rule-based futures strategy — with the sample size, costs, and assumptions shown next to every result.
Examples need no account. A free account includes 5 credits to run your own.
| Company | Insider | Bought | Filed | Value |
|---|---|---|---|---|
| SAICScience Applications International Corp | Cush C. DavidDirector | Sep 22, 2026 | Sep 22, 2026 | $132,500 |
| VFFVillage Farms International, Inc. | DEGIGLIO MICHAEL AChief Executive Officer | Sep 22, 2026 | Sep 22, 2026 | $101,258 |
| RWAYIRunway Growth Finance Corp. | Goldthorpe Edward J.Director | Sep 22, 2026 | Sep 22, 2026 | $42,257 |
| VFFVillage Farms International, Inc. | Woodward Christopher C.Insider | Sep 22, 2026 | Sep 22, 2026 | $30,580 |
| LEN-BLENNAR CORP /NEW/ | BERKSHIRE HATHAWAY INCInsider | Sep 21, 2026 | Sep 21, 2026 | $12,378,038 |
Open-market purchases filed with the SEC through Sep 22, 2026. A snapshot of the real feed, not a mockup. Open the example →
From an idea to the evidence
Discover
See what insiders and Congress are buying
Browse recent open-market insider purchases from SEC Form 4 filings and congressional trade disclosures, with the filing date next to the trade date.
Research
Ask a historical question, get a checkable answer
Ask in plain English what happened after a condition in index futures history. You get the number, the sample size, the definitions, and the code behind it.
Test
Backtest a rule-based strategy
Define entries, stops, and targets, then run them on real futures bars with costs, drawdown, and a held-out period. Stock and ETF backtesting is not available yet.
Congressional trades are also covered under Discover — see how purchases compared with the S&P 500.
Real market data, simulated forward
150 possible paths, resampled from real daily returns.
Hover any line to isolate it. This shows the spread of outcomes baked into recent volatility. It is an illustration, not a forecast.
Micro Nasdaq-100 · simulated 60-day outcomes
Test
See the whole equity curve, costs included.
This is a real run of a textbook strategy — Opening range breakout on MNQ, 5-minute bars — from Sep 6, 2024 to Sep 6, 2026, with commission and slippage. It loses money, and the tool shows that plainly: testing rules before trading them is the point.
- Net profit after costs, win rate, expectancy, profit factor, maximum drawdown
- A held-out later period, so you can see whether results were stable
- Index futures only for now; stocks and ETFs can't be backtested yet
Equity curve, cumulative net P&L by trade (drawdown shaded) — 1 MNQ contract
Net P&L
−$4,260
Win rate
29%
Trades
505
Snapshot generated Sep 24, 2026. Hypothetical results; past performance does not indicate future results.
A real result, open to inspection
After days when the regular session (9:30-16:00 ET) closed down more than 1% from the prior regular-session close, how often did the next regular session close higher than the prior close, and what was the average next-day return?
- Sample
- 335
- days
- % next day up
- 56.7%
- avg next day return
- +0.28%
This looked at MNQ (a Nasdaq-100 futures contract) between June 2019 and September 2026, focusing on the 335 trading days where the regular trading session (9:30am-4pm ET) closed down more than 1% from the prior day's close. Of those 335 sharp down-days, the following regular session closed higher than the prior close about 56.7% of the time - so a bit more often than not, but far from a guarantee. On average across all 335 cases, the next session's return was about +0.28% (0.0028 as a decimal), meaning the typical bounce, when averaged with the losses, was modest. This is purely a description of what happened in this historical window, not a forecast of what will happen after future down days.
- Instrument
- MNQ futures
- Window
- Jun 1, 2019 – Sep 6, 2026
- Snapshot
- Sep 24, 2026
The code that produced it
et = to_et(bars)
rth = rth_session(et)
dates = trading_date(rth)
rth = rth.copy()
rth['date'] = dates
daily_close = rth.groupby('date')['close'].last().sort_index()
daily_ret = daily_close.pct_change()
# days where session closed down more than 1% from prior close
down_days = daily_ret < -0.01
# next day return relative to that day's close
next_ret = daily_ret.shift(-1)
mask = down_days & next_ret.notna()
events = next_ret[mask]
sample_size = int(mask.sum())
if sample_size > 0:
pct_up = float((events > 0).mean())
avg_ret = float(events.mean())
else:
pct_up = None
avg_ret = None
result = {
"sample_size": sample_size,
"pct_next_day_up": pct_up,
"avg_next_day_return": avg_ret
}For context: across all 1,844 regular sessions in the same window, this market closed up on 56.4% of days, averaging 0.09%. A figure like the one at left is a starting point for a question, not a trading edge.
Where the data comes from, and how it is used
Insider filings from 2021-01-01, 17,935 congressional trades disclosed 2012–2026, and futures bars from 2019. Every result states its own window.
Where the data comes from
Insider trades: SEC EDGAR Forms 4 and 4/A. Congressional trades: STOCK Act disclosures collected by Disclosed Capitol. Futures: 1-minute bars from Databento, stitched into continuous series. Stock prices for follow-up returns: EODHD daily prices. Each result shows its own data window.
Dates and timing
Disclosure research is measured from the first trading day after a filing became public — not from the trade date, when nobody else could have acted. Futures times are New York time, and entries in backtests happen on the bar after the signal, never the same bar.
Sample sizes and uncertainty
Repeat purchases of one stock in a month count as one observation, so a single busy company can't inflate the count. Averages come with a confidence interval; when the interval includes zero, the result is described as not distinguishable from no difference.
Costs and what is not modeled
Backtests include per-contract commission and slippage. They do not model partial fills, queue position, taxes, or market impact. Futures series are back-adjusted across contract rolls, so long-horizon percent moves are approximate.
Simple, credit-based pricing
Reading examples and recent-filings feeds is free. Running your own analysis uses credits, and you see the cost first.
Start free
A new account includes 5 credits, and 2 basic insider lookups. Browsing the examples and the recent-filings feeds never needs credits.
That covers about one starter question (5 credits) or a small daily-bars backtest. A standard question is 10 credits.
Pay for what you run
- Historical question10 credits
- Backtest: daily bars, 6 months4 credits
- Backtest: 5-minute bars, 6 months12 credits
- Backtest: 5-minute bars, 1 year18 credits
- Insider deep search10 credits
The cost is shown before you confirm. A failed run is refunded automatically. Purchased credits do not currently expire.
Subscribe if you use it often
- Basic — 100 credits included monthly$29.99/mo
- Plus — 200 credits included monthly$49.99/mo
- Elite — 450 credits included monthly$99.99/mo
Prefer no subscription? Credit packs start at $10 for 25 credits.
Questions
What does TradingQuery cover?
Insider purchases and sales for U.S. stocks (SEC Form 4) and congressional stock trades. Historical questions and strategy backtests currently cover index futures: Micro E-minis (MNQ, MES, MYM) from 2019, and for questions also full-size NQ/ES/YM from 2010. Every tool is labeled with the assets it supports.
Can I backtest stocks or ETFs?
Not yet. The backtester runs on futures price bars only. Stock research today is disclosure-based: what insiders and members of Congress reported buying and selling.
How is an answer checked?
For historical questions, code for your question is written several independent times, run in a sandbox, and reviewed; you see how many attempts agreed, the figures, and the code. Agreement lowers the chance of a coding slip — it does not prove the question was understood the way you meant, so read the definitions.
What happens if there isn't enough data?
If too few observations match, you are told so instead of getting a number, and you are not charged. A failed run is refunded automatically.
Is this investment advice?
No. It reports what happened historically, with its uncertainty. Past results do not predict future ones, and nothing here is a recommendation to buy or sell anything.
Does it place trades?
No. TradingQuery is research software. It does not connect to a broker or execute orders.
Start with an example, then test your own idea.
This tool provides historical statistical information only. It is not investment advice, and past performance does not indicate future results. Trading involves risk of loss.