IC is the Correlation — definition">correlation between a manager's implicit forecasts (revealed by portfolio weights) and subsequent realised returns. Since you can't observe the manager's internal forecasts directly, you infer them from holdings data.
Step 1 — Compute active weights at each period-end
$$aw_{i,t} = w_{i,t}^{\text{fund}} - w_{i,t}^{\text{benchmark}}$$
A positive $aw$ = overweight = implicit "buy" signal. A negative $aw$ = underweight = implicit "sell" signal.
Step 2 — Measure subsequent returns
For each stock $i$, record the 1-month or 3-month return $r_{i,t+1}$ after the portfolio date.
Step 3 — Compute the cross-sectional correlation
At each period $t$, calculate:
$$IC_t = \text{Corr}(aw_{i,t},\ r_{i,t+1}) \quad \text{across all stocks } i$$
Step 4 — Average over time
$$\overline{IC} = \frac{1}{T}\sum_{t=1}^{T} IC_t$$
A consistent positive $\overline{IC}$ over multiple years is strong evidence of genuine skill — not luck.
| $\overline{IC}$ | Interpretation |
|---|---|
| ≈ 0 | No skill — forecasts uncorrelated with outcomes |
| 0.05 | Low but meaningful — still valuable with high breadth |
| 0.10 | Solid skill — approximately the IC of a skilled equity analyst |
| 0.15 | Excellent — very few managers sustain this |
You need monthly holdings disclosures. In India:
For a full holdings-based IC, you'd need the complete monthly portfolio from the fund's AMFI disclosure or a data vendor (Morningstar, PrimeInvestor, etc.).
If full holdings data is unavailable, you can reverse-engineer an implied IC from the fund's observable IR and an estimate of its breadth:
$$IC \approx \frac{IR}{\sqrt{B}}$$
For example: if a mid-cap fund has a 5-year IR of 0.5 and runs ~80 effective independent positions:
$$IC \approx \frac{0.5}{\sqrt{80}} \approx 0.056$$
This is a rough triangulation, not a precise estimate — but it tells you whether the fund's track record is consistent with a plausible level of per-bet skill.
Always benchmark IC against the right index. A mid-cap fund's active weights must be computed vs. the Nifty Midcap 150, not the Nifty 50 — otherwise you're measuring the mid-cap premium as if it were forecasting skill. [10]
Apply this → Once you have a fund's implied IC, plug it into the Fundamental Law with its breadth and TC to see whether its IR is plausible or likely overstated: Explore Funds