Anatomy of a strategy

Three decisions, nothing more

Take away the canvas, the stats, the leaderboard, the follower count. What is left of a Bet2Invest strategy? Three decisions. One sport. One bet selection. One filter tree. That is the whole anatomy, and it fits in one hand.

The part worth tattooing somewhere is the line between the last two. Almost every structural confusion comes from blurring it. The bet selection answers one question: what gets bet? The filter tree answers the other: when?

The bet selection is a fixed instruction. You pick a market (moneyline, spread, total or team total) and a side of it. For a moneyline that means home, away, draw, or one of the market-relative labels, favorite or underdog, resolved from wherever Pinnacle's prices sit on match day (pick'em matchups with no clear favorite are simply skipped). For a spread you also fix the points; for a total, the line and over or under. Whatever you choose, every pick this strategy will ever produce is that exact bet. A strategy never bets "whatever looks good tonight". It bets one thing, every time its conditions hold. Want to bet two things? Build two strategies.

The filter tree holds all the judgment. It is a set of conditions on the matchup: the odds, the league, the teams' recent form, the schedule. Only matchups that satisfy it trigger a pick, always pre-match, never in-play.

Picture a standing order left at a bookmaker's counter. The bet selection is the slip, already filled in. The filter tree is the list of conditions under which the clerk is allowed to hand it over.

The canvas: groups, not a form

The builder does not hand you a form with rows of conditions. It gives you an open canvas with free-floating groups, and each group carries a logical operator that rules everything inside it.

An AND group demands that every condition inside it holds. One miss and the matchup is out. An OR group is happy with any single condition. A NOT group is stricter than it looks: it holds exactly one condition and flips it, so the matchup qualifies precisely when that condition fails. Groups nest freely. An OR tucked inside an AND is the standard way to say "this must be true, and at least one of those".

Now for the rule that catches nearly everyone once: multiple root groups on the canvas combine as an implicit AND. Two separate boxes are not two alternatives. They are two requirements, and a matchup must pass both. If you meant "either scenario", both scenarios belong inside a single OR group.

Drafts are loose, saves are locked

While you build, the canvas autosaves without judging you. Half-finished groups, empty slots, a filter you dragged in and forgot: all of it persists between sessions, none of it is validated. The moment of truth is Save or Preview. That is when the canvas is compiled into the strategy's real configuration and checked. Until then a draft is a sketch. Backtest it, tear it apart, rebuild it. Anyone can create drafts; turning one into a saved, active strategy requires a plan.

Save is a one-way door for the rules. Once saved, the filter tree and the bet selection are locked, and only the description stays editable. The strategy's resource weight (the sum of the computational costs of its filter leaves, counted against your plan's resource points) freezes at the same instant.

The lock is not bureaucracy. A track record only means something if the rules that produced it cannot quietly shift underneath it. Locked configuration is what keeps every strategy's history honest, including yours. When a public strategy shows you its track record, the rules that earned it are exactly the rules you would be following.

A worked example, read back in plain language

Say you build a football strategy. Bet selection: moneyline, underdog. On the canvas, one root AND group holding three things: an odds range condition requiring the underdog's price between 2.50 and 4.00, a league filter narrowing to a handful of leagues you trust the data on, and, nested inside, an OR group with two conditions: the home team is on a no-loss result streak of 3 or more, or the home team's win rate over its last 10 matches is at least 60% (last N result rate).

Read it back and it says: in these leagues, when the underdog is priced between 2.50 and 4.00 and the home side is in form by at least one of two measures, bet the underdog to win.

Every layer earns its place. The odds band keeps you away from heavy longshots and near-favorites alike. The league filter scopes the claim to data you believe. And the nested OR does quiet, important work: demanding both form conditions would strangle the pick count, while accepting either keeps the form requirement meaningful without making qualifying matches vanishingly rare.

One more thing to notice, because it bites: the form conditions watch the home team specifically, whichever side your underdog happens to be. The tree tests exactly what you tell it to test, nothing smarter. Read your own structure back and make sure it says what you meant.

Mistakes that look like bad luck but are structure

The giant single AND is the classic. Stack seven conditions in one group, each individually reasonable, and their intersection can be nearly empty. A backtest that finds 11 bets in four years is usually not a discerning strategy. It is an over-specified one.

The OR built wide is the mirror image. If each branch alone matches half the calendar, the group filters nothing and only pretends to.

NOT has its own trap. Because it inverts a single condition, a NOT wrapped around "odds between 2.50 and 4.00" matches everything outside that band, on both sides at once: longshots and short favorites together, which is rarely what anyone wanted. When a NOT feels necessary, first check whether a positive condition says it more precisely.

Two knobs that survive the lock

A saved strategy still has runtime controls.

The Enabled toggle sits in the strategy header. Switch it to Disabled and the strategy stops scanning upcoming matchups, so no new picks are produced, while the frozen rules stay exactly as they are. Switch it back on and detection resumes. Public strategies cannot be disabled, though, because followers are relying on them.

Odds drift re-checks change when your conditions are evaluated. Normally a matchup is tested at opening odds. With drift re-checks on, detection also re-evaluates when Pinnacle's price moves, so a matchup that missed your odds band at open can qualify later if the market drifts into it.

Where to go next

For the hands-on tour of the canvas itself, read the strategy builder guide. For what each of the 24 filters actually measures, start with the filters overview. And once your tree compiles, the backtest metrics guide explains how to judge what it produces.