TAIGA LABS
Menu

Research Lab

Know what kind of market you are in.

Every report can be filtered by regime, and every cohort answers the same question: what happened next, the last time the market looked like this.

Request accessSign in

Regime

The same read, at five distances.

A market is one thing seen at several ranges. The regime read runs from the month down to the sub-phase, and a coarse call is made of the finer ones underneath it, so you can see the moment a day stops agreeing with its week.

  • Month, week, day, session and sub-phase, on one timeline
  • Directionality, expansion and close location beside the read
  • Every label carries the moment it became known
taigalabs.dev/dashboardDirectionality52%Expansion56%Close location69%Regime, five grainsMonthWeekDaySessionSub-phase Illustrative. Seeded sample data, not a real result.

Cohorts

Sessions like this one, and what they did next.

Pick the conditions that describe today and the cohort finds the sessions that matched. What comes back is a spread of forward paths rather than an average, because the spread is the part you can act on.

  • Match on regime, level, gap, range and session shape
  • Forward paths with their distribution, not a single line
  • Sample size and baseline shown beside every result
taigalabs.dev/research/cohortsSessions like this oneMatching sessions2026-01-172026-06-252026-06-162026-03-282026-08-152026-09-282026-12-252026-02-282026-10-14What happened next Illustrative. Seeded sample data, not a real result.

Watchlist

An edge, and whether it is still there.

Register an edge and it is re-evaluated against every new version of the data. The watchlist keeps the interval it had at registration beside the interval it has now, so drift is something you see rather than something you discover.

  • Effect and confidence interval at registration, kept
  • Re-evaluated as the data is rebuilt
  • Drift flagged against the band it was registered in
taigalabs.dev/research/watchlistRegistered edgesAt registration0.640 [-0.170, 1.330]EvaluatedEffectConfidence intervalRowsDrift9/16, 1:31 PM0.5803,993STABLE9/16, 2:55 AM0.6303,928STABLE9/16, 3:55 PM0.6203,971STABLE9/15, 4:23 AM0.6203,992STABLE9/15, 5:54 PM0.6543,912STABLE9/15, 6:23 AM0.5913,938STABLE9/14, 7:29 PM0.6454,017STABLE9/14, 8:13 AM0.6733,998STABLE9/14, 9:25 PM0.0053,930DRIFTED9/13, 1:15 AM0.6383,910STABLE9/13, 2:33 PM0.5954,014STABLE9/13, 3:23 AM0.6363,980STABLE Illustrative. Seeded sample data, not a real result.

Reports

Analytics you can argue with.

Reports run over a window you choose and show the parameters they ran with. Re-slicing by a facet does not re-run the analysis, so you can take the same result apart several ways without waiting for it again.

  • Session structure, gaps, ranges and opening range
  • Every parameter visible and yours to change
  • Facets re-slice without recomputing
taigalabs.dev/research/reportsOpening range analyticsSymbolNQOpening range15 minTimeframe1mBreak detectionBy wickSessions analysed150sessionsAvg OR range142.9pointsBreakout rate88.0%up and downHold rate76.4%of breakoutsRetracement rate59.2%Closed in OR35.4%Double break51.0%No break2.5%FacetsExclude early closes-10Exclude roll weeks-18Exclude low-density eras-14 Illustrative. Seeded sample data, not a real result.

What a research result is, and is not.

A cohort is a spread

The answer to what happened next is a distribution. Where the spread is wide the report says so rather than reporting its middle.

Sample size is shown

Every result carries how many sessions it rests on and what it is being compared against. A thin sample is labelled thin.

Labels are dated

A regime label records when it became known, so a study cannot quietly use a fact that arrived after the moment it is testing.

Free during the alpha.

We invite traders in small groups. People on the waitlist get in first.

Request access