> ## Documentation Index
> Fetch the complete documentation index at: https://docs.abacuslabs.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Benchmarks

> Which benchmarks Abacus Labs tracks, the names that resolve to each, and how to preview one before you use it.

A `benchmark` on analyze or proposals is what risk and tracking error are measured against. Without one, Abacus Labs uses the account's benchmark, else the S\&P 500.

## What you can send

* **An index name, alias or ticker.** For example `"S&P 500"`, `"SPY"`, `"R1000"`, `"S&P 400"`. `GET /v1/benchmarks` lists them all.
* **Any US-listed ETF with public holdings.** Abacus Labs looks through it to its holdings.
* **A stock**, as a one-name benchmark.
* **An object:** `{"mix": [{"benchmark": "S&P 500", "weight": 0.5}, {"benchmark": "ARKK", "weight": 0.5}]}`, `{"manager": "Pershing Square"}` (a 13F filer's latest public book), or `{"weights": {"AAPL": 0.3, "MSFT": 0.7}}`.

## List the indexes

`GET /v1/benchmarks` costs 1 credit. Each index has its label, the ETF whose SEC N-PORT holdings stand for it, every name and ticker that resolves to it, and the date of its constituents.

```bash theme={null}
curl -s https://api.abacuslabs.ai/v1/benchmarks -H "Authorization: Bearer $ABACUS_API_KEY"
```

```json theme={null}
{ "indexes": [
    { "benchmark": "S&P 500", "index": "$SPX", "proxy_etf": "IVV", "aliases": ["S&P 500", "S&P500", "SP500"],
      "tickers": ["IVV", "SPLG", "SPY", "VOO"], "available": true, "n_constituents": 498, "holdings_as_of": "2026-11-13" },
    { "benchmark": "S&P MidCap 400", "index": "$MID", "proxy_etf": "IJH",
      "aliases": ["S&P 400", "S&P MID CAP 400", "S&P MIDCAP", "S&P MIDCAP 400", "S&P400", "SP400"],
      "tickers": ["IJH", "MDY"], "available": true, "n_constituents": 392, "holdings_as_of": "2026-11-13" },
    ... ],
  "benchmarks": [],
  "note": "Any of these works wherever a benchmark is accepted: ..." }
```

`benchmarks` holds this key's saved benchmarks. `available: false` means this server has no constituents for the index today.

## Preview a benchmark

`POST /v1/benchmarks/resolve` costs 1 credit. It shows what a benchmark resolves to, without using it:

```bash theme={null}
curl -s https://api.abacuslabs.ai/v1/benchmarks/resolve \
  -H "Authorization: Bearer $ABACUS_API_KEY" -H "Content-Type: application/json" -d '{"benchmark": "S&P 400"}'
```

```json theme={null}
{ "key": "IJH", "label": "S&P MidCap 400", "custom": false, "n_constituents": 392, "uncovered_weight": 0.0,
  "components": [ { "kind": "index", "ref": "IJH", "label": "S&P MidCap 400", "source": "SEC N-PORT holdings of IJH",
                    "period_end": "2026-06-30", "as_of": "2026-11-13", "drift_days": 136 } ],
  "top_holdings": [ { "symbol": "P", "weight": 0.0101 }, { "symbol": "TWLO", "weight": 0.0095 }, ... ],
  "as_of": "2026-11-15", "credits_charged": 1 }
```

* `period_end` is the date of the fund's last SEC report. `drift_days` is how far Abacus Labs rolled its weights forward with prices.
* `uncovered_weight` is the part of the benchmark that maps to no security Abacus Labs covers. A benchmark below 50% coverage is refused with `422 invalid_benchmark`.

To measure an account against a benchmark, ask [analyze](https://docs.abacuslabs.ai/concepts/analysis.md), for example: `"What is my tracking error to the S&P 400?"`.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.