The Chang 73 laminated-type shale oil reservoirs (LSOR) in Ordos basin located at the bottom of the source rocks with thickness of 5–20 m, is a lithology complex composed of shale, tuff, silty mudstone, and a small amount of fine sandstone. Shale oil is mainly stored in tuff and silty mudstone with a thickness of centimeters. In seismograms, the Chang 73 shale oil reservoirs have no independent seismic response, because the thickness of the reservoir has not reached the 1/4 wavelength. Besides, the difference in geophysical properties between the reservoirs and surrounding rock is small, thus the reflections of reservoirs are submerged in the strong reflections of the shale strata and difficult to be predicted by conventional seismic inversion method. To solve above problems in unconventional thin reservoirs prediction, we propose a new concept of seismic geobody identifiability (SGI), which indicates the influence of geobody on waveforms of seismic reflection. The SGI is related to the difference of geophysical properties between geobody and surrounding rock, the size of the geological body, the structure of the geobody, and the quality of seismic data. This concept breaks through the limitation of seismic resolution of 1/4 wavelength and provides a theoretical basis for seismic identification of smaller geobodies. The existence of shale oil reservoir makes the seismic reflections of shale have following features: (1) local convexity of the seismic event; (2) widening of seismic seismic wavelet valleys. According to above features, we construct a relationship between widening value of wavelet valley and thickness of reservoirs, and successfully predict shale oil reservoirs above 5 m. The prediction accuracy of 145 validation wells reached 85.5%. The proposed formula realize a accurate prediction of reservoirs, which are submerged in shale seismic strong reflection and have no independent seismic response, and provides a new ideas for predicting unconventional oil and gas reservoirs.