TTradeTech AnalyticsStrategy Research RepositoryRESEARCH ONLY

STRAT-RSI-RB-001 · v1.1

RSI Recovery Breakout

A frozen long-only daily equity signal under institutional research review.

RESEARCHEC-1 — ECONOMIC RESEARCH CANDIDATE SURVIVESNot approved for deployment · confidence MEDIUM
MANDATORY BIAS WARNING
SURVIVORSHIP_BIAS_WARNING_CURRENT_UNIVERSE_RETROSPECTIVE
Current-universe retrospective construction may overstate historical opportunity.
Current result7.33%current-cost counterfactual CAGR
Historical max DD16.65%25% entry budget
Modeled costs₹605,186current-reference +0 bps
Turnover59.02×continuous notional

Executive summary

The strategy seeks a recovery from a prior oversold condition into renewed momentum. v1.0 portfolio construction was rejected after every predefined sizing scenario breached the 30% historical drawdown flag. v1.1 leaves signals unchanged and constrains admission through an entry gross-exposure budget. The 25% budget retains research viability under the current Anand Rathi cost counterfactual; exposure drift, incomplete historical cost schedules and survivorship bias remain material.

25% is an ENTRY GROSS EXPOSURE BUDGET. It is not a hard exposure ceiling. Realized exposure reached approximately 65.9% as existing positions appreciated.

Strategy logic

Entry at close t

  1. RSI[t-1] <= 65
  2. RSI[t] > 65
  3. zero prior RSI crosses above 65 in t-60 through t-1
  4. minimum RSI in t-60 through t-1 is strictly < 30

Execute: next available session open.

Exit at close t

  1. RSI[t-1] >= 40 and RSI[t] < 40
  2. OR Close[t-1] >= EMA200[t-1] and Close[t] < EMA200[t]

Execute: next available session open. If both occur: BOTH.

No profit target, trailing stop, pyramiding, leverage, shorting or averaging.

Parameters

ParameterValueState
RSI period14FROZEN
Entry threshold65FROZEN
Oversold threshold30FROZEN
Recovery lookback60 observationsFROZEN
Exit RSI40FROZEN
EMA period200FROZEN
Position target2%PROVISIONAL for v1.1
Entry exposure budget25% primary; 35% comparatorTESTED / PROVISIONAL
DirectionLong onlyFROZEN

Dataset and semantics

FieldValue
PrimaryRESEARCH_EQUITY_EOD_V1
Requested / physical / accepted500 / 500 / 492
Rows1,839,104
Range1999-01-01 through 2026-09-07
Digestd08db201e5aaf56f4fd4a4bb244433280a1c26594051e67914f5bf96eacc3b4f
BenchmarkRESEARCH_NIFTY_EOD_V1
Price semanticsPROVIDER_STRUCTURAL_ADJUSTED_PRICE
AccountingCONTINUOUS_NOTIONAL
QuarantineEight quarantined instruments excluded; names intentionally not published.

Dhan is the historical data provider. Structurally adjusted OHLC is used directly; no second split or bonus adjustment is applied. Dividends are excluded, so results are price return rather than total return. Merger/demerger economics remain incompletely modeled. Continuous-notional quantities are research coordinates, not claims of exact historical executable share quantities.

Experiment matrix

IDVersionObjectiveVariableValuesResultArtifact
EXP-RSI-001v1.0Original sizing sweepPosition allocation2–20% predefinedREJECTEDresearch artifact RSI-V1-PRIMARY
EXP-RSI-002v1.0Drawdown forensicsExposure/concurrencyAll sizesDF-1 portfolio exposure failureresearch artifact RSI-V1-FORENSICS
EXP-RSI-003v1.1Exposure-budget experimentEntry budget25/35/50/65%25% primary; 35% marginal; 50/65 rejectedresearch artifact RSI-V1-STRESS
EXP-RSI-004v1.1Chronological validationIS/Validation/OOSFixed dates25% remains below 30% historical DDresearch artifact RSI-V1-ECONOMIC
EXP-RSI-005v1.1Moving-block bootstrapPortfolio sequence5/20/60 sessions25% stress P95 DD 24.7–26.7%research artifact RSI-V1-ECONOMIC
EXP-RSI-006v1.1Anand Rathi cost counterfactualCurrent reference costs0/+10/+25/+50 bps25% EC-1; current-cost CAGR 7.33%research artifact RSI-V1-ECONOMIC
EXP-RSI-007v1.1DP tariff sensitivityPer sell eventINR 15/20Separate sensitivity; not confirmed chargeresearch artifact RSI-V1-ECONOMIC

Chronological performance

ANAND RATHI CURRENT-COST COUNTERFACTUAL · NOT HISTORICALLY REALIZED NET PERFORMANCE

SegmentDatesCAGRMax DDSharpeTradesCapturePF
IN-SAMPLE1999-01-01 – 2016-12-315.49%16.65%1.112887.22%5.55
VALIDATION2017-01-01 – 2021-12-318.71%15.27%1.511255.47%1.71
FINAL OOS2022-01-01 – 2026-09-0713.05%15.94%0.98652.32%25.21

Cost sensitivity · 25% entry budget

ScenarioCAGRMax DD
Anand current +0 bps7.33%16.65%
+10 bps7.20%16.73%
+25 bps7.15%16.99%
+50 bps7.00%18.35%

Risk and limitations

MATERIALCURRENT_UNIVERSE_RETROSPECTIVE survivorship bias
MATERIALNo point-in-time historical NIFTY500 membership
MATERIALEntry exposure budget can drift materially; it is not a hard ceiling
MATERIALHistorical transaction-cost schedules incomplete
MATERIALExecution friction not empirically calibrated

Signal-level diagnostic population

CountMeanMedianWin ratePFP10P25P75P90Median MFEMedian MAE
8214+17.0%-1.2%44.9%4.69-12.5%-6.5%+8.8%+46.0%+10.9%-7.1%

Positive skew is diagnostic only; eligible signals were not all economically executable in the portfolio.

Dependence-aware stress

Circular moving-block bootstrap of portfolio daily returns; 500 resamples per block. Fixed seeds: 20300105, 20300120, 20300160.

MODEL-CONDITIONAL STRESS FREQUENCIES — NOT FUTURE PROBABILITIES

BlockP50 DDP75 DDP90 DDP95 DD
5 sessions16.0%18.8%22.6%24.7%
20 sessions16.7%20.3%24.5%26.7%
60 sessions16.4%19.4%22.5%24.7%

Bootstrap frequencies above 20%, 25%, 30%, 35% and 40% DD are conditional on the historical resampling model and are not future-loss probabilities.

Execution costs

ANAND_RATHI_CURRENT_REFERENCE_COST_MODEL · CURRENT BROKER OFFERING / RESEARCH ASSUMPTION

SideModeled total before incremental friction
Buy21.314 bps
Sell19.814 bps
Round trip41.128 bps

Current Anand Rathi cost counterfactual; not historically realized 1999–2026 net performance. Brokerage is 8 bps on each side. Components: brokerage, STT, NSE transaction charge, SEBI charge, stamp duty, GST. DP: Separate tariff sensitivity; not contractually specified. IPFT: UNKNOWN / NOT MODELED.

Risk & limitations

DISCLOSEDCURRENT_UNIVERSE_RETROSPECTIVE survivorship bias
DISCLOSEDNo point-in-time historical NIFTY500 membership
DISCLOSEDEntry exposure budget can drift materially; it is not a hard ceiling
DISCLOSEDHistorical transaction-cost schedules incomplete
DISCLOSEDExecution friction not empirically calibrated
DISCLOSEDDividends excluded; price return, not total return
DISCLOSEDProvider adjustment methodology unavailable locally
DISCLOSEDMerger/demerger economics incomplete
DISCLOSEDDP/IPFT uncertainty
DISCLOSEDContinuous-notional quantities are research coordinates
DISCLOSEDNo live, paper or pilot evidence

Version history

v1.0
REJECTED

REJECTED for portfolio construction

All ten predefined position-sizing scenarios breached the historical 30% drawdown flag.

Change: Original portfolio construction with per-position allocations from 2% to 20%.

Reason: Historical maximum drawdown ranged from approximately 49.4% to 60.7%.

Why this version exists: Establish the frozen strategy's baseline portfolio construction.

What would falsify it: A corrected rerun that keeps every predefined sizing scenario below the 30% drawdown flag.

Lineage: supersedes UNKNOWN — initial version · superseded by v1.1

Evidence added: UNKNOWN — initial version

v1.1
RESEARCH

EC-1 — ECONOMIC RESEARCH CANDIDATE SURVIVES

Signals remain unchanged; portfolio admission uses an entry gross-exposure budget.

Change: Added entry gross-exposure budgets of 25%, 35%, 50% and 65% with a fixed 2% target.

Reason: 25% is primary; 35% is marginal/secondary; 50% and 65% are rejected.

Why this version exists: Address the v1.0 portfolio exposure/concurrency failure without changing signal logic.

What would falsify it: A deterministic rerun, cost model or OOS/stress result that removes the 25% candidate's historical risk headroom.

Lineage: supersedes v1.0 portfolio construction only; signal logic unchanged · superseded by UNKNOWN — current version

Evidence added: Exposure-budget, chronological, bootstrap and current-cost counterfactual evidence.

Institutional verdict

RESEARCH

EC-1 — ECONOMIC RESEARCH CANDIDATE SURVIVES

Continue validation. This repository does not approve buying, investing, deployment or production use.

Next research gate

  1. Authoritative effective-dated Anand Rathi/statutory cost schedules
  2. Empirical execution-friction calibration
  3. Limited one-factor parameter sensitivity
  4. Continued OOS/robustness validation
  5. Point-in-time universe validation if obtainable
AUDITABLE RESEARCH ARTIFACT

Trade evidence · v1.1

This file contains simulated research trades for independent review. It is not a broker execution record. Shadow files, when available, are hypothetical diagnostics and never affect portfolio P&L.

TRADE LOG NOT AVAILABLECompleted portfolio trade logThe retained v1.1 artifact contains metrics and diagnostics but no completed real-trade ledger; no file is published.
TRADE LOG NOT AVAILABLEShadow missed-signal diagnosticNo v1.1 shadow file is published because the retained artifact is not version-specific enough for a safe download.
AVAILABLE · ZIPReviewer packSafe public research materials · SIMULATED / HYPOTHETICALSHA-256 b70eed0ab8729c18…
Download Reviewer Pack