A more recent dataset looked at all penalty shootouts from the World Cup and UEFA European Championships from 1976 to 2016.

What stands out is that for 440 penalties in the database, goalkeepers only remained in the centre of the goal 3% of the time. Over this period, strikers aimed at the centre of the goal more than three times as often as goalkeepers remained central.

Measures, or KPIs, of R&D productivity tend to focus on inputs as simple as numbers of shots versus numbers of goals, and sometimes to break down ‘success’ rates by therapeutic area. So, of the phrases that tend to go unchallenged, but really shouldn’t, ‘shots on goal’ is particularly troubling.

Of course, ‘shots on goal’ is typically used to mean ‘we want a multiple myeloma drug, and we have 4 different kinds of drug that might make it, so we’ll take our shots from different angles.’

In football, the psychology is fascinating. Would a goalkeeper ever want to stand still, maximising the chance of saving a shot, but risking the ire of the crowd who want to see an ‘attempt’ to save? Would a striker feel comfortable to aim for the top corner, knowing that it’s easier for the crowd and teammates to forgive a saved shot than a shot that skies high over the bar? The same is true in pharma. Shooting for the safe areas will be challenged less by the crowd (investors, analysts), even with a failed study, than the creative but risky attempts.

So, let’s use our Multiple Myeloma example. If it is indeed the goal, one could question whether any of the shots are Multiple Myeloma drugs, or broadly haematological cancer drugs, where MM is just seen as the easiest first shot. And if that is the reason, the next question is to ask which assumptions were made. It is a fiercely competitive environment, so the value of hitting the goal might be lower than expected, especially if a traditional late-stage refractory path was seen as a way to get there. Without a significant competitive advantage, any drug launching in Multiple Myeloma today would have a real challenge - and they won’t be judged on how good a drug they are, but how good a label they carry.

This becomes problematic in a portfolio - if each asset has decided to take a shot at Multiple Myeloma, for those same reasons, one might question the hope: that four drugs hit with the same weak label, or that 1 or more die along the way, in which case we’ll never know what might have happened in a different indication. Optimising for risk at an asset level will increase the commercial risk at a portfolio level, even if it has mitigated the technical risk somewhat - the chance that one will hit increases if you take four shots. But that is at the cost of competing for the same trial sites, the same KOLs, the same formulary listing, and of course the cost of the four different (but the same) development plans.

As I have covered before, measuring phase attrition rates often rewards shots instead of goals. Even, in the traditional definition, if the goal is seen as ‘approval’, there is a misleading reward. These surrogates are often individually important - necessary but not sufficient. As in the infamous Brazil-Germany World Cup final, shots matter, and all other things being equal, would be predictive. But all other things are rarely equal.

The phrase ‘shots on goal’ suggests a goal that is so important that we’d better take a lot of shots. But it ignores that the goal itself is a variable, rather than a fixed rectangle. Instead, as in my cartoon, there are some fearsome goalkeepers, preventing shots from getting anywhere near the only meaningful goal (return on invention). Incentivising goals scored (successful launched medicines) is the opportunity - everything else falls into clever analysis, like football’s Expected Goals - interesting but often wide of the mark when acting as a predictor of the outcome of games.

Instead of asking about shots on goal, Asymmetric Learning asks about the goal of the shot. Is it about learning as much as possible, or about scoring? Treating those two shots the same is wrong, and leads to mistaken approaches to R&D.