This study develops a composite Speculation Index (SI) for the Indian equity market using observable trading activity. This index aggregates market signals linked with speculative behavior through five indicators: India VIX, open interest (OI), daily price range, the NIFTY 50 spot-futures spread, and the volume open interest ratio (VOIR). Abnormal movements in these indicators are extracted through Exponential Generalized Autoregressive Conditional Heteroskedasticity (EGARCH) filtering to isolate unexpected fluctuations in trading activity. These abnormal signals form the basis for four SI variants constructed through equal weighting, Principal Component Analysis (PCA), correlation-based weighting, and a Temporal Fusion Transformer (TFT) feature important framework. The study evaluates the indices using Dynamic Conditional Correlation Generalized Autoregressive Conditional Heteroskedasticity (DCC GARCH) estimation, Granger causality tests, wavelet quantile correlation, and wavelet coherence to examine their links with market returns and volatility across different horizons. The results show that periods of stronger speculative pressure coincide with weaker market returns and higher volatility. Among the weighting approaches, the TFT based index produces the most stable weighting structure and displays the strongest and most consistent association with market volatility and returns. SI offers a market based early warning signal of speculative activity that may support regulatory assessments of market conditions and guide investor decisions under changing risk environments.