Rough Volatility
Gatheral, Jaisson and Rosenbaum (2018, QF) further popularise a stream of the literature which emphasises the non-smoothness of volatility paths. These models build on a fractional Brownian motion, with Gatheral et al. proposing a Hurst parameter $H<\frac{1}{2}$ and demonstrating the model's ability to match volatility time series. Fractional volatility models trace back at least as far back as Comte and Renault (1998, MF).
An extensive list of papers in the area is given here. Recent contributions include, for example, El Euch and Rosenbaum (2019, MF) deriving the corresponding characteristic function for the rough Heston model (at least numerically) and Horvath, Jacquier and Tankov (2020, SIAM JFM) studying how rough volatility models apply to the pricing of volatility options.
Recovery Theory
In one of his last published papers, Steve Ross tried the impossible to recover the physical distribution of future stock prices from observed option prices, see his 2015 JF publication.
Jackwerth and Menner (2020, JFE) cast doubt whether the recovery theorem is compatible with future realised returns and variances. Peter Carr answers this question here on Quant.SE and gives an online lecture about the topic here.
Factor Models
Fama and French (2015, JFE) add two new factors to their seminal three factors model (namely RMW and CMA, capturing risk associated to profitability and investment). Hou, Xue and Zhang (2015, RFS) provide an alternative four factor model based on $q$-theory.
Barillas and Shanken (2018, JF) and Stambaug and Yuan (2017, RFS) propose alternative factors. Hou, Mo, Xue, Zhang (2017, RF) show that their $q$-theory model seems to dominate others using spanning regressions.
Other Good Reads
Guyon (2020, Risk) proposes a solution for the joint calibration problem of S&P and VIX options.
Grasselli (2017, MF) presents his 4/2 stochastic volatility model which neatly unifies the Heston model and the 3/2 model.
Zhang (2017, EFM) summarises a large body of research on investment-based asset pricing, culminating in the investment CAPM, which is as simple as the standard (consumption) CAPM yet empirically more successful in explaining the cross-section of stock returns.
Cochrane (2017, RF) gives a recent survey about popular asset pricing models.
RFS editor Itay Goldstein invited leading researchers to share their opinions about which questions will invoke interesting research in asset pricing for the years to come, see the 2021 paper here.
Harvey, Liu and Zhu (2016, RFS) and Hou, Xue and Zhang (2020, RFS) cast doubt over the ever growing ''factor zoo'' in finance by questioning how many ''anomalies'' can robustly be replicated.
J.P. Morgan (2019, SFI) present their extension of Deep hedging using Reinforcement Learning with optimal execution which challenges complete markets and perfect hedging. RL is the new wave in finance industry (trading, execution and portfolio optimisation) and is here to stay, must read