You've presented a betfair scenario which I worked through, albeit there was an incorrect betfair price. I think the main thing you're stuck on is that you keep on comparing the QL with the overround and assuming that is the expected loss which can be quite a flawed assumption at times - more details on this below.
Take a look again at the coin flip example - I had chosen the same odds (1.8) at the bookie for heads and tails to keep it simpler.
If you look back at the post you'll find the overround was 111.1% (or 11.1% depending on convention used) and the vig was exactly 10% (overround-1/overround)
I directly calculated the EV (as the probability is known for a coin flip) - for a £10 bet this was -£1 (and in this case would have been estimated exactly by the vig)
This was (£8 * 0.5) + (-£10 * 0.5) = -£1
I skimped on the details in that post for the sake of space but the amount layed at 2.05 would have been £8.78
As you can see the EV of the lay at the exchange is negative too:
(£8.78 * 0.5) + (-£9.22 * 0.5) = -£0.22
The QL was £1.22, as should be obvious this loss is a combination of the EV of the bet at the bookie (-£1) and the EV of the lay at the exchange -£.22.
That is all the QL is, it is the expected value of both bets combined and realised as a guaranteed loss. Now unless you're getting +ev or 0ev bets at the exchange then your expected value is obviously higher if you don't lay.
I'd hope we both agree that it is obvious from that example that someone not laying there has more value over a series of events
in the long run - he expects to lose on average £1 per event (albeit with lots of variance) vs someone with no variance guaranteeing a loss of £1.22 for each event.
Now you've made an assertion about scenarios where the QL is rather low - I'll agree that that can be true (and indeed the QL can vanish or even reverse - in the case of an arb) however that is generally where you're getting value from a bookie - both the person laying and the person just betting at the bookie benefit from that same scenario!
Consider again the coin example, it might be more informative for you to play around with it yourself - but I'll assert this - if you adjust the bookie odds favouring one of heads or tails while maintaining the same vig you'll find that you can get the same situation - you'll end up with a QL (if you take the favourable bet) lower than the vig.
The flaw in the comparison you're using if you're applying it to these scenarios is then highlighted as the expected value of the favourable bet is low and the vig is no longer a reasonable approximation of it.
We can demonstrate this clearly with the coin flip example:
Lets say the our bookie Paddy lives in an area where most people prefer betting on tails, he's an old school bookie and in order to balance his book and maintain his 10% vig he adjusts his odds a bit more favourably for heads.
So Paddy now has odds of:
Heads: 1.9
Tails: 1.7
(overround here is approximately 111% and vig of approximately 10%)
The betfair price is as before, centred around the true odds, 1.95 : 2.05
So for a bet of £10 on Heads we now want to lay £9.73 at the exchange
Our EV of the bet at the bookie is (£9 * 0.5) + (-£10 * 0.5) = -£0.5
Our EV from laying at the exchange is (£9.27 * 0.5) +(-£9.73 * 0.5) = -£0.23
As you can see again the QL is equal to the sum of the EV from both legs and is -£0.73
This is where the flaw comes in - if you were to compare the QL in that instance to the vig of 10%, you'd be assuming the expected value at the bookie was still -£1, in fact it is -£0.5 in that scenario, half what you're assuming it to be, and less than the QL.
If you're cherry picking low QL bets then you're in effect selecting bets at the bookie that are giving you better value than the vig would suggest ergo it is rather flawed to then cite it and make the comparison between them.
This then brings us back to the point I was making at the beginning - you don't even need to know or need to attempt to estimate that at all in order to make the point about laying as the expected value of the bets at the bookies is the same in either scenario (as it is the same series of bets!). Your EV on that side is the same and thanks to the free bets that is where your money is coming from.
The simple point is just that laying has a cost associated with it, if you chose to lay then you're adding that cost on top of the gains from the bookie in return for getting rid of variance, it doesn't magically affect the EV of the bets at the bookies, that aspect is unchanged, all you're doing is adding a cost in return for removing variance.
(just to note I'm not advocating never laying per say but just trying to get across the point about which gives more value)
Where the confusion may arise is that exchanges display the lay odds actually as back odds. In reality you are accepting someone else backing the event at those odds to you as the bookie.
If you're playing the bookie at the exchange then you're working limit orders and someone else is crossing the spread - ergo your aiming to make +ev lays at betfair (an activity that makes money from the exchange in itself). Generally in matched betting you cross the spread and or pay some commission (if applicable) which makes your activity have -ev.
If you can get perfect matches (which is possible) and laid a whole series of such bets you would retain 100% of your stakes. All that would happen is that in some cases your back bet would win and in some cases your lay bet. The net effect is zero.
Consider roulette. If you back black and red equally you will lose money over time as there is a green space on the wheel. If you could back black and not black you would not lose money.
PS - 3 days in and 500 up - some cheeky extra places helped me out.
As per the coin toss example the same can be illustrated with roulette - you could consider some theoretical free bet roulette offers over 1000 spins where one person bets on red each spin and another person also bets on red each spin but also hedges by betting on black and placing a smaller bet on zero too.
You'll find again that the QL is equal to the EV of all the bets combined. Again illustrating that you're locking in a QL that is larger than the EV of the person just betting on red and not laying.
The only way this is not true is if you've got no cost at all from laying or indeed if your lays have a positive expected value... if they had a positive expected value then you have no need for matched betting and have found a system to make money from betfair.
)