Equilibrium Data Mining and Data Abundance

Abstract

We analyze how computing power and data abundance affect speculators' search for predictors. In our model, speculators search for predictors through trials and optimally stop searching when they find a predictor with a signal-to-noise ratio larger than an endogenous threshold. Greater computing power raises this threshold, and therefore price informativeness, by reducing search costs. In contrast, data abundance can reduce this threshold because (i) it intensifies competition among speculators and (ii) it increases the average number of trials to find a predictor. In the former (latter) case, price informativeness increases (decreases) with data abundance. We derive implications of these effects for the distribution of asset managers' skills and trading profits.

Machine Learning and the Stock Market

“…First, our results show that the technical trading could have been profitable given the availability of computational power. This is evident in our results for early years of applying our machine learning algorithm, and it is in line with Dugast and Foucault's (2020) theoretical finding that better computing power raises the average quality of predictors and hence profitability.…”

“…It is beyond the scope of this article to attempt to disentangle the functional form of the time trend. 10 The decrease in out-of-sample alphas is consistent with the theoretical findings of Dugast and Foucault (2020). They show that the increase in computing power raises i) average quality of predictors and ii) price informativeness.…”

“…The phenomenal increase in computational power is necessary for our article's analysis. The profitability of our algorithm in earlier years (e.g., the 1970s), when current computing power was barely even imaginable, matches the first effect in Dugast and Foucault (2020) the increase in the quality of predictors (and subsequently, trading profits). Later on, as the computing power grows even more (post-2000), we observe the increase in price informativeness that leads to a decrease in trading profit.…”

“…The idea is that, because of increased computing power, "arbitrageurs' ability to attack mispricing has improved over time" (Linnainmaa and Roberts (2018)). On this note, Dugast and Foucault (2020) provide a theoretical study showing that "greater computing power raises the average quality of the predictors used in equilibrium and therefore price informativeness. The first effect raises speculators' expected trading profit while the second reduces it."…”

Fintech: what’s old, what’s new?

“…The mixed empirical evidence is echoed by recent theoretical models of technological progress and information acquisition. Dugast and Foucault (2020) show that advances in computing power increase market efficiency by reducing the cost of searching for signals. However, data abundance can have the opposite effect, because investors must now search among many potential signals and may ultimately find it more costly to acquire information (the "needle in the haystack" effect).…”