THREAD: Using LGOP data to assess single game performances
Our metrics are designed to uncover the long term underlying talents of players around the world, but here is a snapshot into how we can help measure how well a given game went (1)
As our example, we will be pitting @Arsenal 2003/04 vs @ChelseaFC 2004/05. Here are the two sides, along with their respective positional scores. Note that for this simulation, I'll be using IS (2)
Here we have the game data after 75000 simulations. We can see the average possession, chances and ultimately, goals that each side had. We can then see the likelihood of each result occurring, and the expected points for both teams (3)
So what can we do from this information.
Firstly, I get that possession isn't the best stat BUT it does have some value in predicting the 'dominance' you could expect either side to have. In this instance, you could imagine that Chelsea would by and large have the edge (4)
Moving on to shot actions, using these chance and goal figures, we can benchmark a pre game xG, both for and against, that we might envisage given the relative strength of both sides. This is brilliant for both managing realistic.. (5)
...expectations going into a given game and also reflect back on past performances. Football is a game of such fine margins that every small gain can end up making a big difference. We believe the simulation capabilities offered... (6)
... by our system PAREJO deliver an objective method for both pre and post game analysis by highlighting strong and weak areas for each side, as well as assessing the most likely outcomes of games so coaches can plan accordingly (7)
This is just a snapshot into the game analysis opportunities offered by LGOP. If you have any further questions on this thread, or would like to know how we can help your team gain on-pitch edges, please get in touch.
Feedback and shares much appreciated!!!!
You can follow @lgopfelix.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: