The Funnel
The full sell-side process on one page: market → region → sector → stock → spread → size. Score the macro, get a regime verdict, build the pair, size it properly, and the page writes the note for you. Everything runs in your browser — nothing is sent anywhere.
? How to use thisTen minutes the first time, two minutes a month after that. Read once, then skip. Guide
The rhythm
This is not a page you refresh all day. It runs on the release calendar, and the release calendar is monthly.
| When | What you do | How long |
|---|---|---|
| First weekend of the month | Re-score all fourteen checkpoints from the latest releases. Set the cycle state. Generate the note and keep it. | ~30 min |
| Each Saturday | Update the spread on any live pair. Leave stage 1 alone unless something big printed. | ~10 min |
| Per new idea | Stages 3 and 4 only. Stage 1 already told you whether you are allowed to be long. | ~5 min |
| Daily | Glance at the VIX. If it has moved 25% in either direction, come back and re-run stage 4. | ~1 min |
Stage 1 — score the market
Fourteen checkpoints, each set to bullish, neutral or bearish. The grey line under each one tells you the threshold, so you are not guessing. Take the readings from the primary release, not from a headline about the release — the headline is usually the noisiest number in the report.
- Score first, opinion second. The whole point is to set the checkpoints before you decide what you think. If you find yourself adjusting one because the total looks wrong, you have stopped using the tool and started arguing with it.
- Neutral is a real answer. A reading that is genuinely ambiguous should be neutral. Forcing everything to bullish or bearish makes the score swing on noise.
- Read past the headline. Durable goods is the standing example: the headline fell 4.5% because of aircraft orders while the ex-transport figure rose 1.3%. One of those is signal.
- A flip costs two points. Moving a checkpoint from bullish to bearish removes a bull and adds a bear. Scores move faster than they look like they should.
Stage 2 — pick the tilt
Set the cycle state from the score plus the direction of the last two PMI prints. Rising and above 50 is not the same as falling and above 50, and the difference is the entire trade.
If the score and the cycle state disagree, the page says so in the note. Do not proceed past that warning — one of the two inputs is stale, and sizing on a contradiction is how people end up long into a rotation.
Stage 3 — build the pair
A pair expresses a relative view with the market risk taken out. You need both legs in the same currency unless you intend to be long that currency too — long a London line against a New York line makes you a currency trader whether you meant to or not.
- Short more dollars of the lower-beta leg. That is what the ratio is for. Matching dollar amounts leaves you accidentally long the market.
- Reject a ratio above 2. At that gap you are not hedged, you are levered, and a market fall will hurt you even if you were right about the relative call.
- The spread is the position. Chart the ratio, not the legs. The exit is the ratio crossing back through its own mean, whatever the macro is doing.
- Three ideas for every position. Track the spread weekly and only fund the ones that are consistently working. Most ideas should die on the watchlist, and that is the system doing its job.
Stage 4 — size it
Volatility sets the stop; the stop and your edge set the size. In that order.
- Be honest about the win rate. This is the input people fake. Use a number from a trade log, not from memory. If you have no log, start at 30% and find out.
- Bet half Kelly. Full Kelly is the growth-maximising bet if your inputs are exactly right, and your inputs are never exactly right. Half Kelly gives up a little return for a much shallower drawdown.
- Stops are hard, targets are soft. The stop protects you and should be obeyed mechanically. The target is where you start rolling the stop up, not where you must sell.
- The VIX rule. A 25% rise means portfolio risk has roughly doubled — halve the book. A 25% fall means it has halved. This is arithmetic, not a forecast.
1 The marketFourteen checkpoints. Are we in an expansion, and should you be long at all? —
Set each checkpoint from the latest release. Prefilled with the readings as of 6 August 2026. Bullish +1, neutral 0, bearish −1.
Thresholds: +5 or better full risk-on, long cyclicals against defensives. +1 to +4 long bias with hedges. 0 or below flat or defensive. −3 or worse net short. The score is a permission slip, not a forecast.
2 The sectorIf the market is going up, which parts beat it? Otherwise just buy the index. —
Where the cycle sits determines the tilt. This is the rotation rule, not a view.
The single most useful signal in the whole framework is the first PMI decline after a year-plus of expansion. It is early, it is unambiguous, and it is the moment to rotate rather than to sell.
3 The pairExpress the view with the market risk hedged out. Beta-matched, both legs. —
Short more dollars of the lower-beta leg so both sides move together with the market. What is left is the relative call, which is the only thing you actually had a view on.
4 The sizeVolatility sets the stop. The stop and the edge set the size. —
5 The noteEverything above, written out. Copy it into your journal. —
Not investment advice, and not a signal service. This is a calculator: it turns readings you supply into the arithmetic of a well-known top-down process. It does not know what will happen, and a high score has been wrong before. Every number on this page comes from you or from your own reading of a public release. Nothing is stored and nothing is sent anywhere — refresh the page and it is gone.