Deducting scores from a base score

This seems like an overly complex scoring strategy. Can you describe your use-case in more depth? It might help us brainstorm some alternatives for you. 

Hi Owen,

Thank for the reply. We are working with a potential client who wants to have an examen with a base score of 10. Everytime a student answers wrong we want to deduct 0,4. Meaning after 25 wrong answers the score will be 0. (Eventhough there are still 15 questions more)

We want to show the client a score of > 55 when he has passed and < 55 if he has failed. If I would translate this to a linair scoring it would mean the following:

A student will start to earn points after 15 correct questions. 16th being 0,4, 17th being 0,8 etc.

This would mean that at question 29 the student would have passed the exam with a score of 5.6.  

My current solution is to set the pass % on 29/40 = 73% But that is a very odd number for both the client and student instead of the >5.5. 

Sorry if this is confusing, i'm trying to explain as good as possible.

How important is it that the actual score be passed to an LMS?  If the final score is less important than a pass/fail on the condition X percent or X score is achieved... 
You could use a variable to track their score and then answer a hidden T/F question to generate the results slide and pass a complete/passed incomplete/failed response to the lms.

You could even add a slider that adjusts upward or downward from a starting point of 10 as a feedback mechanism for the user.

The score being passed to an LMS is a must unfortunately. Do you have an example of the slider as a feedback mechanism. This actually sounds like a viable option. 

Thanks so much for helping out!

Here is the biggest challenge with this type of scoring:
When a question adds to a score when correct and subtracts from a score when incorrect, you are essentially doubling the penalty to the user for incorrect answers. 

Think of it this way.  Assume I have 10 questions worth 10/each or costing -10 when incorrect.
If a user misses the first question, it takes another question just to get them back to a break even score of 0. That leaves 8 questions left to score so their highest potential score after missing 1 question is 80, even with a 90% answer accuracy. 

Sometimes, it's ok to question the client to consider why they want something and recommend something different. You don't sell someone a chainsaw when all they need is a pocket knife just because they asked for it.