Epistemic status: high uncertainty.
Even if I'm wrong this model could be useful to others, you can plug your own numbers. Please correct my assumptions in the comments if you think they're wrong or you have new/better data.
Chance of infection
We don't know what percentage of people will get infected. Some experts say 40-70%. Around 50% of cases are symptomatic, but for young people it's closer to 90%. Asymptomatic cases seem not to be infectious. The rate of infection among young people is being underreported because the symptoms are milder and not detected as covid but are mistaken for common flu. According to the stats, the chance of infection is very small for children and 4x lower for 20yo than 60yo.  This is wrong, but we can accept it and move on because having mild symptoms is not fatal and hence equivalent to not having covid at all. From now on, by "infection" I mean "noticable infection". The chance of being asymptomatic is already factored in. Note: say 100% of people get exposed and the risk of infection grows linearly from 0yo to 60yo, that would give us total of about 50-70% of an infected population. Best effort estimate of infection chance:
infection = 0.4
Lack of treatment
We don’t know what percentage of people who have coronavirus die. That’s because most of infected people so far had received hospital care. Once the pandemic spreads, most people will not receive hospital care because hospitals will be full. What would the mortality rate for untreated patients be?
"every demographic has approximately equal hospitalization rates, which other sources suggest are 15% to 20%."
Those "demographics" are crude estimates, but still 0-49 had the same admittance rate as older. With regards to those 20% who require hospitalization:
"5% of people who are diagnosed with Covid require artificial respiration. Another 15% need to breathe in highly concentrated oxygen - and not just for a few days."
Some studies show 6% of people in critical condition, while still 18% of people have a severe case. It's not clear what exactly "severe" means. The best I can estimate with that is:
base_rate_untreated_mortality = 0.12
Adjusting for demographics
How does age affect that? The data on that is messy but seems like <50yo have around 3x smaller chance of a "critical condition". The chance is 2% for 15-49, 2.5% for people of all ages with no preexisting condition, 6% overall. Note: those people received treatment, so this is a lower bound. What happens in completely untreated cases is still unknown.
Age (30yo) seems to lower the chance by 1.5x
age = 1/1.5
Having no preexisting conditions seems to lower the chance by 2x.
npec = 1/2
Note: age and having no preexisting conditions are correlated. In this model, together they offer a protection of 3x.
Being male increases the chance by 1.5x.
sex = 1.5
No data controlled both for age and for preexisting conditions. There is some probability that the quadrant of (young, no-preexisting condition) have very low mortality. No idea what that probability is so I'm going to say 50%.
quadrant = 0.5
There is a chance that the virus will mutate into low-mortality strain. High amount of infected in Germany and still very low deaths - indicates that this may already happened.
mutate = 0.1
The "cure" which disables transmission may be found, and manufactured in enough quantities. 
cure = 0.1
Climate may be an important factor in slowing the disease down.
climate = 0.3
Even if (young, no-preexisting condition) receive preferential treatment, the hospital systems seem to break down at 5K infected in an area of around 20M people (Italy, South Korea). Inside that area, the total number of infected may be 10M, even with some containment at the peak of infection it may be 4M, implying 800K people needing hospitalization, implying 1.25% percent of getting treatment if infected at that time. Oxygen tanks could get depleted at some point. Also, we may run out of doctors. Due to high viral load, doctors are under larger chance of being infected:
"A high number of medical workers have been infected — 10% in the Lombardy region in the north, where the virus first appeared."
Still, containment measures may work well, so the hospitals will not be overcrowded, which implies that treatment will be available.
treatment = 0.2
context = (1 - quadrant) * (1 - mutate) * (1 - cure) * (1 - climate)
age30_treated_mortality = 0.002
risk = (base_rate_untreated_mortality * age * sex * npec * (1 - treatment) + age30_treated_mortality * treatment) * context * infection
As steve2153 says, infection chance and treatment chance both depend on which scenario we are in: population-wide infection vs small-size infection. Therefore the model should be simpler:
risk = (base_rate_untreated_mortality * age * sex * npec) * context * infection