Everything you need to get from an idea to a tested, ranked hypothesis —
and to make sense of what Whalur tells you along the way. Each section
has its own link, so you can send a colleague straight to the part they need.
A hypothesis is a testable claim about market behaviour, written in plain
language. Open New hypothesis from the
Hypotheses page and describe the idea the way
you would explain it to a colleague — for example
“after a large block trade prints at the bid, the stock tends to
drift lower over the next two sessions.”
A useful hypothesis names three things:
A trigger. The observable event that starts a trade — a whale move, a gap, a volume spike, a close above a level.
A direction and a horizon. Which way you expect price to go, and over how many bars, minutes or days.
An exit. When the trade ends: a fixed holding period, a target, a stop, or the next opposite signal.
Whalur turns the text into a formal strategy specification before it runs
anything. If the description is missing one of those three parts you will be
asked to fill it in rather than getting a vague result. Keep one idea per
hypothesis — two mechanisms in a single test make the verdict impossible to
attribute.
Hypotheses are public by default and appear in the library
and Rankings. Paid plans can mark a hypothesis
private, star it for quick access and share it with friends.
Running a test and choosing daily, minute or tick data
Press Run on a hypothesis page and Whalur backtests the
specification across the research universe of roughly 60 liquid tickers using
real Alpaca OHLCV and trade data. Runs are queued; a daily run usually finishes
in under a minute, while minute and tick runs take longer because there is far
more data to scan. The page updates on its own — you do not need to refresh.
Pick the data granularity that matches the horizon of the idea:
Daily. One bar per session. Best for swing ideas measured in days or weeks, and the fastest to run. Start here unless your trigger genuinely happens inside the day.
Minute. One-minute bars. Use for intraday setups — opening-range breaks, momentum in the first hour, fades into the close. Slower, and a single session already contains 390 bars, so keep the holding period explicit.
Tick. Individual trade prints. This is the only granularity that can see a block trade — a whale move — as it happens, so any hypothesis triggered by large prints needs it. The report shows a funnel of raw prints scanned → block trades found → entries taken so you can see how selective the trigger was. Tick data is the most expensive to run and is metered on paid tiers (see Billing and tiers).
Every run is kept in the hypothesis’s run archive, so you can
compare a daily run with a minute run of the same idea, or re-run after a
change to the specification and see whether the verdict moved. Parameter
sweeps run the idea across a grid of settings and show the evidence behind each
cell.
Whalur does not score a strategy by win rate. A strategy that wins 80% of the
time and loses everything on the other 20% is a bad strategy. The report
measures three things instead, and every report opens with a highlighted
in layman’s terms translation of what they mean together.
Expected value (EV). The average profit or loss per trade,
after every winner and loser is counted, shown with a confidence interval.
A positive EV whose interval stays above zero is the thing you are looking
for; a positive EV whose interval straddles zero may simply be noise.
The pooled EV on a hypothesis averages across all of its runs
so one lucky test cannot carry it.
Consistency. How evenly the edge shows up across tickers
and across time. An idea that works on two stocks in one quarter and
nowhere else will earn a Low consistency flag even if its
headline EV is positive.
Trade frequency. How often the trigger actually fires.
A beautiful edge that appears twice a year is hard to act on; the
Low sample flag warns you when there are too few trades to trust
the numbers.
The verdict badge at the top of the report combines those into one assessment:
Positive EV — the edge is real across the sample: this is a PASSED hypothesis.
Weak positive EV — positive on average but thin, inconsistent or under-sampled. Treat as INCONCLUSIVE and refine the trigger or extend the data.
No edge — the trades average out to roughly zero. FAILED: the market does not behave the way the hypothesis claims.
Negative EV — the trades lose money on average. Also FAILED, but sometimes worth inverting: an idea that reliably loses is an idea that reliably works the other way.
No trades — the trigger never fired. Loosen the condition or switch to a finer granularity.
Rankings order hypotheses by how much of the universe has been tested first,
then by pooled EV — so a fully-tested idea always sits above a partly-tested
one with a flashier number.
A signal is a passed hypothesis firing in live data: the
trigger just happened on a real ticker, and Whalur publishes the call with the
evidence behind it. The Signals page lists every
signal with its status — open, closed, expired — and you can narrow it by
status or by date range. Each signal has its own page with the whale move that
triggered it and the hypothesis it came from.
Alert subscriptions tell Whalur to notify you when a signal
fires, so you are not refreshing a list. From Alerts
choose what you want to hear about — a specific hypothesis, a ticker, or every
new signal — and how:
Email to the address on your account.
SMS to a verified mobile number.
Push notifications in the Whalur mobile app.
Immediate alerts go out the moment a signal is generated. If you would rather
have one message a day, switch a subscription to the daily digest,
which bundles everything that fired since the last one. Pause or delete a
subscription from the same page at any time. Signals are research, not
instructions: the tradable call is shown to members, but whether and how to act
on it is yours to decide.
Whale Watch is the daily issue: the largest
block trades of the session, which tickers the whales were leaning on, and any
signals those moves triggered. It is written from the same tick data the
backtests use, so what you read there is exactly what a tick-granularity
hypothesis would have seen.
Whale moves is the full archive — every published block trade, searchable by ticker.
The weekly roundup collects the week’s moves and the signals they produced in one place.
Members can have Whale Watch delivered by email each morning; turn it on from Alerts.
A good way to use it: when a move in Whale Watch catches your eye, open the
ticker’s history, form a hypothesis about what tends to happen after moves like
it, and test it on tick data. That loop — observe, hypothesise, test — is what
Whalur is for.
Whalur has three plans; the full comparison is on Pricing.
Free — $0. One active hypothesis at a time with the full backtest engine behind it, the public library (top-ranked rows withheld), mechanisms, signals and alerts. No card required.
Basic — $29 / month. Unlimited hypotheses, private hypotheses, the full ranking table, 20 AI strategy builds a month, and friends to share research with.
Pro — $99 / month. Everything in Basic plus unlimited strategy builds, a priority backtest queue, premium tick data and the creator marketplace — sell access to your private research and keep 70%.
Subscriptions are billed monthly through PayPal. Upgrade from
Pricing or from Account → Billing,
where every charge is listed with its amount, date and what it was for. Usage
counters — hypotheses in play, strategy builds used this month — sit on the
same tab so you can see how close you are to a limit before you hit it.
Cancelling takes two clicks on the Billing tab. It stops the
next renewal; paid features stay available until the end of the period you have
paid for, then the account drops back to Free. Nothing is deleted — private
hypotheses become read-only until you upgrade again, and unlock exactly as you
left them. To close the account entirely, or to download everything you have
made as JSON, use the Export and Danger-zone controls on
Account → Settings.
