Mercedes strategy has not been up to scratch of late and to me they seem to be too computer simulation based on their thinking , no real gut feeling . If Max does not drive as per the simulation ,ie faster the strategy is wrecked ! I wonder if they have race drivers on the strategy team or is it all boffins ?
The area that caused the greatest problems for predicting the best race strategy at the French GP and actually pretty much all GPs is the tyre degradation. In earlier sessions on Friday the cars will have conducted long runs to provide data for the race strategy models. This data has the greatest variance and causes the greatest problems for the strategists. So they look at what their competitors have done and their own data and populate their mathematical model of the race with tyre degradation data with some variance, along with other relevant data.
During the race the core race model is used with live data being fed into it and a subset of simulations run to predict the race outcome and suggest alternative strategies to change the race outcome in your favour. As the live data is more accurate than the pre race data, it quickly becomes clearer what is more likely to happen. This data is used with input from the tyre engineers to help the strategists make decisions.
I am not aware of any team using an ex-driver to help with race strategy decisions live during a race and I would not want them involved. The race strategists have far more experience of how races are predicted to turn out and actually turn out. And with more accurate tyre data from the tyre engineers and the mathematical predictions they will make better predictions over the whole season. An ex race driver will be doing pattern matching from a much smaller subset, without all the additional info, i.e. creating their own race model in their head. Whereas the race strategists create far more accurate models and hence over the season will make better decisions.
Some historical info to put this all into perspective.
A good friend of mine developed the first stochastic pre race strategy model in 1997 - 98 whist at McLaren. That was used very effectively for many years. In 2004 (I had left McLaren in 98 but he stayed there for many years) he had taken the core race model and linked it to the live data to make live predictions of what would happen. He then left and went to RBR and then Ferrari, developing better solutions each time. He now works as an Operational Research consultant for the odd motorsports team but also for companies outside of F1. Prior to the French GP he sent me a correct prediction of the top 3...
In 2009 James Vowles (I employed him in 2004 to develop the BAR Honda race strategy toolset and apply it during race events) was in charge of strategy at Brawn GP. Throughout the season the decisions were spot on. Whilst we had the fastest car for the first 4 races, with no money to develop the car, other teams caught up and overtook Brawn. So much so that by the second half the Brawn was on average 4th fastest. But very good reliability, poor decisions by other teams, Jenson starting well and then recovering his loss of form near the end of the season and very good strategic decisions, resulted in Brawn winning both championships.
Back to now:
James Vowles is the strategic director at Mercedes and so the same person is making the big calls on the pitwall. There were logical reasons for everything that they did but mistakes were made. The 1 stop should have won them the race, but as I have stated already, killing your front tyres, which already had graining problems, by driving too close to VER, meant that both BOT and Lewis were not able to defend at the end of the race, and they should have been able to keep VER behind them.
So as I have stated, the biggest mistakes were made by the tyre engineers at Merc and the drivers, who destroyed their tyres far too early in their long stint.