> For the complete documentation index, see [llms.txt](https://algoalpha.gitbook.io/algoalpha-user-manual/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://algoalpha.gitbook.io/algoalpha-user-manual/atlas-ai-backtesting-agent/backtesting-configurations.md).

# Backtesting Configurations

Atlas uses fixed, published settings so performance numbers are comparable across the pre-computed strategy search universe.

{% hint style="warning" %}
Strategy search and the weekly batch database support AlgoAlpha SSA, ILPAC, and MC (Momentum Concepts) signals.
{% endhint %}

#### Coverage (instruments and time)

| Setting                          | Value                                                                                                                                                                     |
| -------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Asset universe**               | **85** liquid symbols across **stocks, ETFs, crypto, forex, commodities, and futures** (exact list evolves with the platform)                                             |
| **Bar timeframes**               | **5m**, **15m**, **1h**                                                                                                                                                   |
| **Indicators (search universe)** | **SSA**, **ILPAC**, and **MC** — all three of AlgoAlpha's premium signal products are searchable.                                                                         |
| **Strategy shape (search)**      | **1 trigger** plus **0–2 state filters/states**; each combo is tested in **long**, **short**, and relevant **both** directions where applicable. See Signal combinations. |

#### Indicator settings

Atlas evaluates SSA, ILPAC, and MC using their default indicator settings. The internal parameters of these indicators are not optimized or varied between backtests.

When reproducing an Atlas strategy in TradingView, keep the indicators at their default settings for the closest comparison. Changing an indicator's settings may produce signals and performance results that differ from those shown in Atlas.

#### History length and data window

| Setting           | Value                                                                                     |
| ----------------- | ----------------------------------------------------------------------------------------- |
| **Bars per test** | **20,000** bars of OHLCV history on the chosen timeframe for each pre-computed result     |
| **End of window** | Aligns with the **most recent** data available when that result was produced or refreshed |

So a "20,000 bar" test on **1h** covers a longer wall-clock span than on **5m**, but the **number of bars** is the same.

#### Costs and capital (standardized)

| Setting                          | Value                                                                                                         |
| -------------------------------- | ------------------------------------------------------------------------------------------------------------- |
| **Commission**                   | **0.1% of trade value per fill**, applied on **entries and exits**                                            |
| **Starting capital (reference)** | **$100,000** notional account for the standardized metrics you see in search and in typical dashboard reports |

Slippage and margin are **not** simulated for simplicity

#### How metrics are scaled

**Sharpe ratio** and similar risk metrics use **timeframe-appropriate annualization** (shorter bars imply more periods per trading year). You do not need the exact factors to interpret results—only that **5m, 15m, and 1h** numbers are scaled consistently within each timeframe.

Search results come from a **batch refresh** of the strategy database:

* **Cadence:** **Weekly** full refresh so rankings track updated history.
* **Typical schedule:** Operations often target **Monday 00:00 UTC** for the pipeline run; the exact window can move for maintenance or holidays.
* **Effect for you:** Metrics and rankings for searchable strategies can **change week to week** as new data is ingested. A strategy that ranked first last week is not guaranteed to rank the same after the next refresh.

Between refreshes, results are **stable** for that build of the database.

Strategies you've starred in chat are re-checked against each weekly rebuild — see Saved Strategies for how the Current / Deprecated status is decided.

#### Execution model (plain language)

For the **pre-computed** library, each tested strategy follows a **simple position model**: at most **one direction at a time**, entries and exits driven by the trigger and state rules for that combo, evaluated on the same **close-based** bar sequence used for all peers. That keeps thousands of results **mutually comparable**.

The entry/exit conditions for all strategies are **mirrored** for the opposite direction.

#### Asset List

**Cryptocurrencies**

BTCUSDT, ETHUSDT, BNBUSDT

**Stocks (NYSE)**

BABA, BA, BAC, C, CVX, DIS, GE, GM, GME, GS, HD, IBM, JNJ, JPM, KO, LMT, MA, MCD, MMM, MRK, NKE, ORCL, PFE, PG, SNOW, T, UBER, UNH, UPS, V, VZ, WMT, XOM

**ETFs (AMEX / ARCA)**

ARKK, DIA, EEM, EFA, GDX, GDXJ, GLD, HYG, IWM, QQQ, SLV, SPY, TLT, USO, VTI, XLE, XLF, XLK, XLV

**Forex**

AUDJPY, AUDUSD, EURAUD, EURGBP, EURUSD, EURJPY, GBPAUD, GBPJPY, GBPUSD, NZDUSD, USDCAD, USDCHF, USDJPY

**Commodities**

XAUUSD, XAGUSD, UKOIL

**Futures**

NQ, ES, RTY, YM, ZC, ZW, ZS, GC, SI, CL, NG, PL, 6E, 6J