With only low-throttle, no-ignition, the only net torque comes mainly from friction and pumping - you can estimate this as appx. 1 bar for a 4-stroke NA engine at no-throttle. Then, the torque can be calculated from T=MEP*V_displacement/(2*PI*N_c), where N_c is the number of strokes per power cycle (2 for 4-stroke, 1 for 2-stroke). So, on a 5.0 liter NA gasoline engine, you would expect about (100,000 Pa)*(0.005 m^3) / (2*3.1415*2) = 39.8 N-m (29.4 lb-ft) of torque, probably lower since 1 bar FMEP/PMEP combined figure is likely high for those operating conditions. You can modify the numbers using estimates from your application, but you won't get very far 'braking' by this method --> a majority of the work which could help you (compressing intake gas) is lost during the expansion stroke, since the cylinder acts as an air-spring.
What ivymike alluded to with the diesels & the exhaust/compression braking effect - ie, open the exhaust valve at TDC after the compression stroke - would add significantly more torque, since all the compression work is lost. But, this requires an external mechanism (ala Jake Brake or similar).