A problem is that statistical "expectation" answers (percentage chance x value) are not a value one may expect to receive. They are a value which on average may be expected if there were repeated events.
For a lawyer, with many clients and cases, an accurate expectation will suggest how they can maximise their overall case wins.
But its a lot less clear how valid that is for a single client with a single case. An analogy might be - roll a die. If the number is 3 - 6 you double your money, 1 or 2 you lose it all. Now, the expectation is that for every pound/dollar you bet, you'll receive a gain of 33% (bet 6 times => lose 2 units, win back 8 units => long term expected gain is 8/6 x your bet value). Now, I as your friend suggest that as you have an expectation of profit, you should borrow heavily and bet your shirt and all you can on a single roll of the dice, and enjoy the outcome! You probably don't want to. Whereas if I suggested betting many times for £100 each, you'd probably like it, because the expectation (long term) will have a very good chance to play off.
With a single case having a 70% chance of a win and 30% chance of a lose, the risk isn't of a gradual gain/gradual loss. You get a single event only, which is that you roll the dice once, and a significant chance of both major gain and major loss. Expectations and trends and averages just aren't a good way to decide it. It can guide, but it shouldn't decide.
A better frame of mind is this. Here are 3 medical situations (not great analogies, but a high-impact single event choice we are all familiar with):
- Suppose you had to consider whether or not to have surgery to fix a disability (representing the damage you're currently living with in the legal situation). The surgery is guaranteed to be quick and pain-free, so the only question is the likely outcome. You have a 60% chance of complete success but 40% chance of making it worse.
- As above, but this time there's an 85% chance of success - but also a 15% chance of death.
- Finally, suppose you had usually-fatal-in-18-months cancer. Chemotherapy would be painful and cost all you have, but gives you a 20% chance of complete cure, a 20% chance of 2 more years of life, and a 60% chance of completely wasting your money and being painful as well.
Can you see from these 3 examples, how in single event cases, many people might reject an 85% success option, or accept a painful and costly 20-40% chance. Different people with different views will often differ on what they'd do. They probably would not just do a "statistical working" to decide between the options, though.
Tl;dr - expected gain/loss is a helpful hint, but a bad decider.
In a single case, you need to engage your intuition about the other party, and your own needs and attitudes.
You ask for a rule of thumb, but for a single case, the best rule of thumb is to ask, "what do you think is likely to happen, what are the risks and impact if it works/fails, and what are your attitudes to those".
Update + worked example:
A better solution is the negotiator's mindset, which is very similar. You will need to put yourself into the other side's position, and ask how they will see it.
How much is it worth to them, to settle it prior to more expenses, or to avoid whatever risks they may face if it goes to a full court hearing? If you won, how much would a judge (of in some cases jury) feel is reasonable for you to have asked?
Especially in the UK (and perhaps there are similar rules in other countries: check!), you may be required to have made a reasonable attempt to settle before the hearing. If the amount you asked is not seen by the court as "reasonable", or is more than your eventual win, then you might be required to pay the legal costs the other side was forced to incur as a result of you seeking an inflated settlement.
This is a calculation where the 70/30 expectation is not the best guidance. Let's try an example, with lots of data:
Suppose that we ask our lawyer what advice he thinks the other side is getting from their legal advisor. Your lawyer's guess is that they've been told that the damages for written matters (100k) are 50/50, the damages for verbal and past practice matters are 60/40 in their favour, and if they settle now, you don't seem to be claiming for (say) 50k of emotional distress or other things, but in court you'll probably get 15k if any part of your main case wins,and of course they'll have to pay 30k more of fees if it proceeds that far.
Their client now considers. You seem determined but perhaps could be coerced to pay less if they flash some money. Their legal insurance may not cover cover all of it if it goes to court. The lawyer will have told them that generally they don't want to go to a court case if they have less than 65-70% chance of a win.
If it goes to court, the cost of your win, if any, will be compounded by 15k + 30k = 45k. Plus management time, too. Their (guessed) advice is therefore that there's a good chance of a substantial loss (50%), and if it did go to court that loss could easily be quite a lot bigger. They have a good incentive to try and push it to the line, and settle before that lot can happen. You might want to be pushy, and point out the extra claim amounts they face in a claim, if they don't settle, and suggest that an offer of 130k is generous and "aimed at avoiding legal costs for us both".
On the other hand let's change the nature of the case, but not its chances. Suppose the advice they are likely to have been given is that you won't be likely to win more than 40k. However the claim relates to something disreputable that would come out in court (maybe you were a whistleblower and they sidelined you and forced you to resign, after you tried to report some shady practices, or the case relates to sexism/racism/tolerance of bad or harassing behaviour, or boosting invoices and borderline accounting fraud), and you're pretty sure they would lose at least Sizeable Client X and probably suffer commercial and reputational backlash costing 600k revenues over the next 3 years whether they win or lose.
Now the case odds are far more favourable to them and far less favourable to you, but the downside of going to a hearing is far more severe for them, win or lose. You might seek 300k damages, on the basis that "this is the claim I am prepared to accept, for full and final settlement of all breaches, any emotional damage, and all costs".
You anticipate they will tell you it's unreasonable and you will not win that sum, but you also think if you stick to your view, they'll finally offer close to it at the last moment, on the morning of the hearing.
If they don't offer to settle by the start of hearing, you'll offer 250k as you walk in, but you would rather go to court than offer less, even if a court win was only likely to be 150k, because you think they'll settle at some point for over 200-225k, maybe after they see you are determined to see it through, at the end of day 1.
Last, if they have a legal expenses insurer, the insurer will have many cases. Their goal will purely be to minimise their total loss, with no emotional investment, which is different again.
So when it's just one case, these are the kinds of factors to consider.
Remember that like any negotiation, a settlement is worth whatever they and you agree. If it's worth a lot to them, you may in some cases decide to ask them for a lot, because even though costly, its still their cheapest/best option. If you can't settle, then you have to ask what your best alternative is. Do you want the risk of it going to court, and how sure are you?