What Cramer Holds vs What He Recommends: Signal-Time Features in 16,701 Mad Money Recommendations (2018-2024)

Abstract

We study 16,701 long stock recommendations from Jim Cramer's CNBC Mad Money (January 2018-December 2024) and ask which signal-time features and prior-disclosure states separate the recommendations that subsequently beat the S&P 500 from those that don't. Forward returns are measured at one-, three-, six-, and twelve-month horizons against the same-window return of the SPY ETF. Because Cramer frequently returns to the same names-NVDA alone is recommended 179 times-we first group repeated mentions into 5,459 continuous position sequences, then label each sequence by whether it contains an on-air Charitable Trust ownership disclosure, advisory hold language, or neither. Four signal-time features and sequence-level labels jointly structure forward performance in the 2018-2024 sample. Market capitalization, trailing 90-day return, and VIX regime are directly observable at recommendation time; engagement tier is used descriptively at the sequence level and operationally only when Cramer has disclosed ownership or hold advice in a prior episode. The sharpest single result is that the casual-buy underperformance is almost entirely a small-cap phenomenon. In the small-cap casualbuy cell, the unconditional one-year stock return is-11.9% versus SPY at +12.7%; outside crisis periods (VIX < 30), the corresponding market-neutral pair trade earns +28.7% per position (p < 0.0001) and remains profitable after typical short-selling costs. Beyond this we show that (i) the engagement-tier ordering portfolio > hold-recommendation > casual buy holds directionally and survives controls for size, momentum, and sector, (ii) Cramer's timing on his own Trust holdings is performance-dependent-he is more likely to mention a portfolio name after a recent run-up, and his return mentions on alreadyowned names after ≥15% prior drawdowns predict +24.7% one-year alpha, a pattern that is absent on similarly-drawn-down stocks he does not own, and (iii) the small-cap pair trade flips sign in crisis regimes (VIX ≥ 30), so the regime distinction is essential.

Keywords: Jim Cramer, Mad Money, stock recommendations, retail investor attention, recommendation return predictability

JEL Classification: G11, G12, G14, G17, G40