By Arizmic LLC14 min readPublished 2026-08-10Last reviewed 2026-08-07

Reference

KPI, Result, and Artifact Dictionary

Look up canonical metrics, applicability, Result tabs, record types, artifacts, exports, and units.

App path

Results -> selected Run Output
Studies -> Study Results
Docs -> Reference -> KPI, Result, and artifact dictionary

Read the identity before the number

Every KPI belongs to one immutable Run Output or Study Result. Before comparing a value, confirm the subject version, data and window, parameter or policy values, execution model, requested and effective fidelity, capital and cost assumptions, analysis depth, and retained-detail profile.

The same metric name can answer a different question when those inputs differ. A rounded card is a summary, not a replacement for the exact retained record.

Result record dictionary

Record	Meaning	Opens in	Contains
Strategy Run	One standalone execution of one exact Strategy version and parameter vector	Results	One Run Output with applicable metrics, trades, execution evidence, validation, and artifacts
Strategy Retained Run	One Strategy candidate retained or promoted from a Study	Results	The candidate's Run Output plus its parent Study and candidate identity
Strategy Optimization	One Strategy Study that compared candidates or trials	Studies	Leaderboard, candidate/trial accounting, parameter evidence, folds, robustness, and retained Run Outputs
Portfolio Run	One coordinated execution of one exact Portfolio version and policy	Results	Account-level Run Output, member attribution, admission, exposure, policy, trades, and artifacts
Portfolio Retained Run	One Portfolio candidate retained or promoted from a Study	Results	Coordinated Run Output plus its parent Study and candidate identity
Portfolio Optimization	One Portfolio Study that compared coordinated candidates	Studies	Candidate set, objective and feasibility evidence, admission/policy comparisons, and retained Run Outputs

A Run Output explains one execution. A Study Result explains a search, comparison, or selection. Do not reconstruct a candidate denominator from one promoted Run Output.

Evidence-state dictionary

Public state	Meaning	How to treat the value
Available	The required retained inputs were read and the named calculation or record applies.	Interpret it with its unit, denominator, fidelity, and limitations.
Not applicable	The component does not apply to this subject, method, fidelity, or retained decision.	Do not replace it with zero or treat it as a failure.
Blocked	Missing, malformed, stale, insufficient, unsupported, or unreadable evidence prevents the claim.	Follow the named blocker and corrective action; do not infer the value.
Not retained	The run completed, but its declared detail profile did not keep that optional output.	Rerun with the required retention only if the detail is necessary.
No trades / insufficient sample	The run is valid but the denominator is empty or too small for the metric.	Keep valid zero counts separate from ratios or statistics that cannot be calculated.

An absent value, a zero, and Not applicable are different facts. A zero profit with ten trades is not the same as a run with no trades; a missing Sharpe is not a Sharpe of zero.

Capital, profit, and return KPIs

KPI	Canonical meaning	Unit	Applicability and boundary
Starting Capital	Capital base supplied to the recorded execution model	Account currency	Requires a defined capital convention; not inferred from ending capital
Ending Capital	Starting capital plus the cash-flow and P&L effects included by the run's account convention	Account currency	Read with deposits, withdrawals, open positions, and realized/unrealized treatment
Gross P&L / Gross Total Profit	Profit before the modeled commission, fee, slippage, and other retained cost drag	Account currency	Available only when gross and cost fields are retained consistently
Net P&L / Total Net Profit	Profit after the costs included by the execution model	Account currency	The primary profit figure for reconciliation; its cost scope must be visible
Gross Profit	Sum of positive closed-trade net P&L values	Account currency	Used with Gross Loss for Profit Factor; it is not the same as gross-before-cost P&L
Gross Loss	Absolute sum of negative closed-trade net P&L values	Account currency	Reported as a nonnegative loss magnitude
Total Return	Net P&L divided by the applicable starting-capital base	Percent	Requires positive defined starting capital and consistent cash-flow treatment
Annualized Return	Return scaled to the run's declared annualization convention	Percent per year	Available only with sufficient duration and a valid calendar/sampling basis
Cost Drag	Difference attributable to retained commissions, fees, slippage, and other named costs	Account currency, and percent where shown	Never includes an unmodeled cost merely because it exists in live trading
Break-even Cost	Additional declared cost at which the selected outcome reaches its break-even condition	Price, currency, basis points, or per-trade unit shown	Robust evidence only; nonlinear sizing or execution can make it Not applicable

For a normal closed-trade reconciliation:

CODE

net P&L = gross-before-cost P&L - retained cost drag
ending capital = starting capital + included net cash-flow effects

The displayed run contract is authoritative when open P&L, financing, deposits, withdrawals, multiple currencies, or other cash flows apply.

Drawdown and path KPIs

KPI	Canonical meaning	Unit	Interpretation boundary
Maximum Drawdown	Largest peak-to-later-trough decline in the retained equity path	Account currency	Depends on the equity sampling path and included cash flows
Maximum Drawdown %	Maximum drawdown divided by the applicable preceding equity peak	Percent	Requires a positive valid peak; not interchangeable with loss divided by starting capital
Drawdown Duration	Time or observations spent below a prior equity peak	Bars, events, sessions, or elapsed time shown	An unrecovered drawdown is censored at the end of the run, not falsely marked recovered
Time to Recovery	Observations from drawdown start until the prior peak is regained	Bars or time shown	Unavailable for an unrecovered episode
Return / Drawdown	Recorded return or net-profit measure divided by the matching maximum-drawdown measure	Ratio	Numerator and denominator units must match the displayed contract
Volatility	Dispersion of the retained return series under the declared sampling and annualization convention	Percent or return units	Sampling frequency, gaps, and annualization must be present

Equity and drawdown are path measures. They cannot be reconstructed reliably from only the first and last capital values.

Closed-trade KPIs

KPI	Canonical meaning	Unit	Required denominator or input
Trade Count	Number of retained completed trade records under the run's trade-matching convention	Count	Exact trade schema and matching policy
Wins / Losses / Scratches	Closed trades with positive, negative, or zero net P&L	Count	Net trade P&L
Win Rate	Winning closed trades divided by all closed trades, including scratches in the denominator	Percent	At least one closed trade
Loss Rate / Scratch Rate	Losing or zero-P&L closed trades divided by all closed trades	Percent	At least one closed trade
Average Trade / Expectancy	Total closed-trade net P&L divided by closed-trade count	Account currency per trade	At least one closed trade
Profit Factor	Gross Profit divided by Gross Loss	Ratio	Positive Gross Loss; an undefined denominator is not an infinite value by default
Largest Win / Loss	Highest or lowest retained closed-trade net P&L	Account currency	Exact closed-trade records
Holding Period	Duration between the trade's retained entry and exit boundary	Bars, events, or elapsed time shown	Depends on the execution clock and trade-matching convention
Trade Gap	Distance between consecutive retained trade boundaries defined by the metric	Bars or time shown	Requires ordered compatible trades
Long / Short Split	Counts and contribution grouped by retained trade side	Count, currency, or percent	Requires an unambiguous side field
MAE	Maximum adverse excursion while a trade was open	Ticks, price, currency, percent, or R as labeled	Requires a retained intratrade path at sufficient fidelity
MFE	Maximum favorable excursion while a trade was open	Ticks, price, currency, percent, or R as labeled	Requires a retained intratrade path at sufficient fidelity

Tables are paginated. Always check the total row count and page controls before treating visible rows as the full sample.

Risk-normalized and risk-adjusted KPIs

KPI	Canonical meaning	Unit	Availability rule
R-multiple	One trade's net P&L divided by its exact retained initial risk amount	R	Not applicable when a valid risk amount was not recorded
Average R / Expectancy R	Mean of available closed-trade R-multiples	R per trade	Requires comparable valid R values; missing-risk trades remain disclosed
SQN	Square root of the R-observation count multiplied by mean R, divided by sample standard deviation of R	Ratio	Requires at least two valid R values and nonzero dispersion
Sharpe	Mean excess return divided by return volatility under the displayed return series, benchmark, frequency, and annualization convention	Ratio	Degenerate, insufficient, or undefined return series remain unavailable
Sortino	Return relative to downside deviation under the displayed target and sampling convention	Ratio	Requires enough downside observations and an explicit target/basis
Calmar	The run contract's return measure divided by maximum drawdown percent on the same scale	Ratio	Requires positive maximum drawdown percent; inspect whether the numerator is total or annualized return
Probabilistic Sharpe	Estimated probability that the observed Sharpe exceeds the named benchmark after accounting for sample shape	Probability	Robust evidence with sufficient observations, skew, and kurtosis inputs
Deflated Sharpe	Sharpe evidence adjusted for the number and distribution of compared trials	Probability or ratio shown	Study-selection evidence; requires a valid multi-trial denominator
Minimum Track Record	Estimated observation count needed to exceed the named Sharpe benchmark at the declared confidence	Count	Requires a valid observed Sharpe and distribution assumptions

Risk-adjusted ratios are not portable when the return sampling, risk-free or benchmark rate, annualization, or treatment of gaps differs. Preserve those fields with every export or comparison.

Portfolio KPI dictionary

KPI	Meaning	Normal unit or denominator
Gross Exposure	Total absolute long and short exposure at the coordinated account boundary	Notional currency or percent of the displayed capital/margin base
Net Exposure	Signed long exposure minus short exposure	Notional currency or percent of the displayed base
Open Risk	Loss budget currently exposed under the Portfolio's recorded risk convention	Account currency or percent of risk/capital budget
Capital / Margin Utilization	Amount of the named capital or margin resource in use relative to its limit	Percent with exact denominator
Turnover	Traded activity relative to the Portfolio's declared capital or notional basis	Ratio or percent per displayed period
Concentration	Share of exposure, risk, or contribution held by the largest named member, instrument, bucket, or category	Percent with category and denominator
Participation Rate	Eligible member or candidate participation under the recorded policy	Percent of the named eligible set
Admission Rate	Admitted intents divided by the declared eligible/requested intent set	Percent; clipped, suppressed, and rejected categories remain separate
Member Contribution	Portion of Portfolio P&L, cost, exposure, risk, or drawdown attributed to one immutable Strategy Configuration	Currency, percent, or risk unit shown
Risk-rule Violations	Retained events where a named Portfolio constraint reached its violation condition	Count with rule and severity
Risk Scaling	Multiplier or limit adjustment applied by a named Control or regime rule	Multiplier, percent, or limit unit shown

Member contributions reconcile under shared capital, overlap, allocation, and cost rules. Do not sum percentages or standalone member returns unless the Portfolio result explicitly defines that operation.

Robustness and scenario evidence

Evidence	What is retained	What it does not mean
Bootstrap interval	Interval from declared resampling unit, iteration count, seed, and confidence level	A guaranteed future range
Monte Carlo distribution	Simulated paths or outcome distribution under the recorded transformation	A market forecast
Matched-random / null comparison	Observed result compared with a declared null or randomized baseline	Proof that all alternative explanations were removed
Parameter sensitivity	Outcomes around retained neighboring parameter values	Stability outside the tested neighborhood
Temporal stability	Rolling, fold, session, or period variation	Untouched out-of-sample evidence after repeated inspection
Cost sensitivity	Outcome changes under declared cost changes	Live fill or cost certification
Benchmark-relative evidence	Alpha, beta, tracking error, information ratio, or capture measures under an exact aligned benchmark	Availability without compatible benchmark currency, calendar, and timestamps
Portfolio dependence	Aligned member dependence, diversification, concentration, or risk contribution	Independence of members or stability after membership changes
Counterfactual / stress scenario	Result under one precisely declared policy or scenario change	Evidence about scenarios that were not run

Each component carries its own state. There is no hidden blended “quality score” that can override a blocked component.

Result tab dictionary

Tab	Primary question	Subject
Overview	What ran, what happened, and what limits the conclusion?	Strategy and Portfolio
Validation	Are identity, data, fidelity, execution, and rerun evidence sufficient?	Strategy and Portfolio
Performance	How do profit, costs, path, distributions, and contribution reconcile?	Strategy and Portfolio
Chart	Where did path, positions, and retained executions occur?	When Interactive Chart was retained
Parameters	Which exact Strategy parameters and bindings produced this run, and can they be reused?	Strategy
Portfolio Policy	Which composition, priority, allocation, capital, admission, and risk rules governed the run?	Portfolio
Strategy Attribution	Which immutable member Configurations contributed P&L, cost, exposure, and risk?	Portfolio
Risk & Exposure	How were shared capital, exposure, margin, and risk limits used?	Portfolio
Admission & Conflicts	Why was member intent admitted, clipped, netted, suppressed, or rejected?	Portfolio
Positions & Trades	How did intent become orders, fills, positions, and closed trades?	Strategy and Portfolio
Monte Carlo & Robustness	How sensitive is this retained run to the declared sample, path, cost, time, or scenario changes?	When requested evidence applies
Outputs	Which exact files and bounded event records were retained?	Strategy and Portfolio
Diagnostics	How did the run spend time/resources, produce artifacts, or fail?	Strategy and Portfolio

Artifact dictionary

Artifact role	Typical contents	Reader or action
Run Output / evidence manifest	Immutable identity, lineage, assumptions, statuses, schema versions, checksums, and references	Overview, Validation, Outputs
Performance summary	Point metrics and applicable compact path summaries	Overview and Performance
Return / equity / drawdown series	Ordered observations used for path and risk evidence	Performance, Chart, Robustness, Notebook
Trades	Closed-trade summaries, P&L, costs, duration, R, MAE/MFE, and lineage where retained	Positions & Trades, Performance, Notebook
Execution events / event journal	Intent, order lifecycle, fills, position changes, costs, queue outcomes, and stable sequence	Positions & Trades, Outputs, Validation
Chart data	Bars or events required by the Interactive Chart profile	Chart
Execution model	Fill, ordering, cost, latency, queue, liquidity, and fidelity assumptions plus fingerprint	Validation and Diagnostics
Replay integrity	Coverage, anchors, gaps, resets, epochs, ordering, and L2/L3 eligibility	Validation, Outputs, Diagnostics
Robustness package	Resampling, Monte Carlo, return-strength, sensitivity, and scenario outputs	Monte Carlo & Robustness
Portfolio policy and decision evidence	Candidate policy, admission/conflict records, attribution, exposure, risk, counterfactuals, and scenarios	Portfolio-specific tabs
Diagnostics	Phase timings, counts, resources, warnings, errors, and publication/recovery detail	Diagnostics
Study candidate/trial evidence	Search space, candidates/trials, objective, pruning, failures, folds, selection, and retained-output links	Strategy or Portfolio Study Results

An artifact path is only a locator. Its role, schema, checksum, status, and Run Output or Study Result identity determine whether it is usable evidence.

Export and Notebook actions

Copy preserves the selected identity or safe path for a review note.
Reveal opens only an allowed, present artifact inside its governed storage root.
Export Result snapshot packages the selected identity, validation state, lineage, and requested diagnostic references for support or independent review.
Open in Notebook binds the exact result and artifact to a Notebook investigation without mutating the source Result.

An exported file or Notebook calculation does not become built-in Result evidence merely because it was derived from a valid artifact. Preserve the source identity, schema, checksum, units, and calculation method.

How Arizmic processes result evidence

CODE

immutable execution request and resolved lineage
  -> candidate-local orders, fills, positions, trades, and account path
  -> compact metrics calculated before ranking and retention
  -> Study candidate/trial selection and declared finalist retention
  -> applicable Summary or Robust evidence
  -> schema and checksum validation
  -> immutable Run Output or Study Result publication
  -> bounded readers, charts, exports, and Notebook handoff

Metrics are derived from exact retained inputs before display formatting. Rounded cards and charts are readers of that evidence; they are not the source of truth. Optional detailed artifacts may be retained only for declared finalists, while the complete candidate, trial, fold, failure, cancellation, and pruning denominator stays with the Study Result.

If a reader cannot prove the required schema, input, denominator, or unit, it returns an explicit unavailable state instead of guessing a value from nearby files or the latest product version.

Start with Results Overview, then use Validation before interpreting Performance. Use Positions & Trades for exact execution, Monte Carlo & Robustness for retained sensitivity evidence, and Outputs for artifact identity and integrity.

Read the identity before the number

Result record dictionary

Evidence-state dictionary

Capital, profit, and return KPIs

Drawdown and path KPIs

Closed-trade KPIs

Risk-normalized and risk-adjusted KPIs

Portfolio KPI dictionary

Robustness and scenario evidence

Result tab dictionary

Artifact dictionary

Export and Notebook actions

How Arizmic processes result evidence

Next