diff --git a/examples/basic.rs b/examples/basic.rs
index 86df4f5..76e7776 100644
--- a/examples/basic.rs
+++ b/examples/basic.rs
@@ -1,7 +1,7 @@
 use kickscore as ks;
 
 fn main() {
-    let mut model = ks::BinaryModel {};
+    let mut model = ks::BinaryModel::new(ks::BinaryModelObservation::Probit);
 
     // Spike's skill does not change over time.
     let k_spike = ks::kernel::Constant::new(0.5);
@@ -16,9 +16,9 @@ fn main() {
     ];
 
     // Now we are ready to add the items in the model.
-    model.add_item("Spike", k_spike);
-    model.add_item("Tom", k_tom);
-    model.add_item("Jerry", k_jerry);
+    model.add_item("Spike", Box::new(k_spike));
+    model.add_item("Tom", Box::new(k_tom));
+    model.add_item("Jerry", Box::new(k_jerry));
 
     // At first, Jerry beats Tom a couple of times.
     model.observe(&["Jerry"], &["Tom"], 0.0);
@@ -35,20 +35,20 @@ fn main() {
     model.fit(true);
 
     // We can predict a future outcome...
-    let (p_win, p_los) = model.probabilities(&[&"Jerry"], &[&"Tom"], 4.0);
+    let (p_win, _p_los) = model.probabilities(&[&"Jerry"], &[&"Tom"], 4.0);
     println!(
         "Chances that Jerry beats Tom at t = 4.0: {:.1}%",
         100.0 * p_win
     );
 
     // ... or simulate what could have happened in the past.
-    let (p_win, p_los) = model.probabilities(&[&"Jerry"], &[&"Tom"], 2.0);
+    let (p_win, _p_los) = model.probabilities(&[&"Jerry"], &[&"Tom"], 2.0);
     println!(
         "Chances that Jerry beats Tom at t = 2.0: {:.1}%",
         100.0 * p_win
     );
 
-    let (p_win, p_los) = model.probabilities(&[&"Jerry"], &[&"Tom"], -1.0);
+    let (p_win, _p_los) = model.probabilities(&[&"Jerry"], &[&"Tom"], -1.0);
     println!(
         "Chances that Jerry beats Tom at t = -1.0: {:.1}%",
         100.0 * p_win
diff --git a/src/fitter.rs b/src/fitter.rs
new file mode 100644
index 0000000..8d76303
--- /dev/null
+++ b/src/fitter.rs
@@ -0,0 +1,7 @@
+mod batch;
+
+pub use batch::BatchFitter;
+
+pub trait Fitter {
+    fn add_sample(&mut self, t: f64) -> usize;
+}
diff --git a/src/fitter/batch.rs b/src/fitter/batch.rs
new file mode 100644
index 0000000..3f55f22
--- /dev/null
+++ b/src/fitter/batch.rs
@@ -0,0 +1,40 @@
+use crate::kernel::Kernel;
+
+use super::Fitter;
+
+pub struct BatchFitter {
+    ts_new: Vec<f64>,
+    kernel: Box<dyn Kernel>,
+    ts: Vec<usize>,
+    ms: Vec<usize>,
+    vs: Vec<usize>,
+    ns: Vec<usize>,
+    xs: Vec<usize>,
+    is_fitted: bool,
+}
+
+impl BatchFitter {
+    pub fn new(kernel: Box<dyn Kernel>) -> Self {
+        BatchFitter {
+            ts_new: Vec::new(),
+            kernel,
+            ts: Vec::new(),
+            ms: Vec::new(),
+            vs: Vec::new(),
+            ns: Vec::new(),
+            xs: Vec::new(),
+            is_fitted: true,
+        }
+    }
+}
+
+impl Fitter for BatchFitter {
+    fn add_sample(&mut self, t: f64) -> usize {
+        let idx = self.ts.len() + self.ts_new.len();
+
+        self.ts_new.push(t);
+        self.is_fitted = false;
+
+        idx
+    }
+}
diff --git a/src/item.rs b/src/item.rs
new file mode 100644
index 0000000..c7b892b
--- /dev/null
+++ b/src/item.rs
@@ -0,0 +1,11 @@
+use crate::fitter::Fitter;
+
+pub struct Item {
+    pub fitter: Box<dyn Fitter>,
+}
+
+impl Item {
+    pub fn new(fitter: Box<dyn Fitter>) -> Self {
+        Item { fitter }
+    }
+}
diff --git a/src/lib.rs b/src/lib.rs
index 5ea3605..bcb0555 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -1,6 +1,10 @@
 // https://github.com/lucasmaystre/kickscore/tree/master/kickscore
+mod fitter;
+mod item;
 pub mod kernel;
 mod model;
+pub mod observation;
+mod storage;
 
 pub use kernel::Kernel;
 pub use model::*;
diff --git a/src/model.rs b/src/model.rs
index ec81ec3..5e486f2 100644
--- a/src/model.rs
+++ b/src/model.rs
@@ -1,23 +1,91 @@
+use std::f64;
+
+use crate::fitter::BatchFitter;
+use crate::item::Item;
 use crate::kernel::Kernel;
+use crate::observation::*;
+use crate::storage::Storage;
+
+pub enum BinaryModelObservation {
+    Probit,
+    Logit,
+}
 
 pub struct BinaryModel {
-    //
+    storage: Storage,
+    last_t: f64,
+    win_obs: BinaryModelObservation,
+    observations: Vec<Box<dyn Observation>>,
 }
 
 impl BinaryModel {
-    pub fn add_item(&mut self, name: &str, kernel: impl Kernel) {
-        //
+    pub fn new(win_obs: BinaryModelObservation) -> Self {
+        BinaryModel {
+            storage: Storage::new(),
+            last_t: f64::NEG_INFINITY,
+            win_obs,
+            observations: Vec::new(),
+        }
+    }
+
+    pub fn add_item(&mut self, name: &str, kernel: Box<dyn Kernel>) {
+        if self.storage.contains_key(name) {
+            // raise ValueError("item '{}' already added".format(name))
+        }
+
+        self.storage.insert(
+            name.to_string(),
+            Item::new(Box::new(BatchFitter::new(kernel))),
+        );
     }
 
     pub fn observe(&mut self, winners: &[&str], losers: &[&str], t: f64) {
-        //
+        if t < self.last_t {
+            // raise ValueError("observations must be added in chronological order")
+        }
+
+        let mut elems = self.process_items(winners, 1.0);
+        elems.extend(self.process_items(losers, -1.0));
+
+        let obs: Box<dyn Observation> = match self.win_obs {
+            BinaryModelObservation::Probit => {
+                Box::new(ProbitWinObservation::new(&mut self.storage, &elems, t, 0.0))
+            }
+            BinaryModelObservation::Logit => {
+                Box::new(LogitWinObservation::new(&mut self.storage, &elems, t, 0.0))
+            }
+        };
+
+        self.observations.push(obs);
+
+        for (item, _) in elems {
+            // item.link_observation(obs)
+        }
+
+        self.last_t = t;
     }
 
     pub fn fit(&mut self, verbose: bool) {
         //
     }
 
-    pub fn probabilities(&mut self, team1: &[&str], team2: &[&str], t: f64) -> (f64, f64) {
+    pub fn probabilities(&mut self, team_1: &[&str], team_2: &[&str], t: f64) -> (f64, f64) {
         (0.0, 0.0)
     }
+
+    fn process_items(&self, items: &[&str], sign: f64) -> Vec<(usize, f64)> {
+        /*
+        if isinstance(items, dict):
+            return [(self.item[k], sign * float(v)) for k, v in items.items()]
+        if isinstance(items, list) or isinstance(items, tuple):
+            return [(self.item[k], sign) for k in items]
+        else:
+            raise ValueError("items should be a list, a tuple or a dict")
+        */
+
+        items
+            .iter()
+            .map(|key| (self.storage.get_id(&key), sign))
+            .collect()
+    }
 }
diff --git a/src/observation.rs b/src/observation.rs
new file mode 100644
index 0000000..ffadc72
--- /dev/null
+++ b/src/observation.rs
@@ -0,0 +1,7 @@
+mod ordinal;
+
+pub use ordinal::*;
+
+pub trait Observation {
+    //
+}
diff --git a/src/observation/ordinal.rs b/src/observation/ordinal.rs
new file mode 100644
index 0000000..58651c8
--- /dev/null
+++ b/src/observation/ordinal.rs
@@ -0,0 +1,58 @@
+use crate::storage::Storage;
+
+use super::Observation;
+
+pub struct ProbitWinObservation {
+    m: usize,
+    items: Vec<usize>,
+    coeffs: Vec<f64>,
+    indices: Vec<usize>,
+    ns_cav: Vec<f64>,
+    xs_cav: Vec<f64>,
+    t: f64,
+    logpart: usize,
+    exp_ll: usize,
+    margin: f64,
+}
+
+impl ProbitWinObservation {
+    pub fn new(storage: &mut Storage, elems: &[(usize, f64)], t: f64, margin: f64) -> Self {
+        /*
+        assert len(elems) > 0, "need at least one item per observation"
+        */
+
+        ProbitWinObservation {
+            m: elems.len(),
+            items: elems.iter().map(|(id, _)| id).cloned().collect(),
+            coeffs: elems.iter().map(|(_, sign)| sign).cloned().collect(),
+            indices: elems
+                .iter()
+                .map(|(id, _)| storage.get_item(*id).fitter.add_sample(t))
+                .collect(),
+            ns_cav: Vec::new(),
+            xs_cav: Vec::new(),
+            t,
+            logpart: 0,
+            exp_ll: 0,
+            margin,
+        }
+    }
+}
+
+impl Observation for ProbitWinObservation {
+    //
+}
+
+pub struct LogitWinObservation {
+    //
+}
+
+impl LogitWinObservation {
+    pub fn new(storage: &mut Storage, elems: &[(usize, f64)], t: f64, margin: f64) -> Self {
+        LogitWinObservation {}
+    }
+}
+
+impl Observation for LogitWinObservation {
+    //
+}
diff --git a/src/storage.rs b/src/storage.rs
new file mode 100644
index 0000000..5fe7dec
--- /dev/null
+++ b/src/storage.rs
@@ -0,0 +1,36 @@
+use std::collections::HashMap;
+
+use crate::item::Item;
+
+pub struct Storage {
+    keys: HashMap<String, usize>,
+    items: Vec<Item>,
+}
+
+impl Storage {
+    pub fn new() -> Self {
+        Storage {
+            keys: HashMap::new(),
+            items: Vec::new(),
+        }
+    }
+
+    pub fn contains_key(&self, key: &str) -> bool {
+        self.keys.contains_key(key)
+    }
+
+    pub fn insert(&mut self, key: String, item: Item) {
+        let index = self.items.len();
+
+        self.items.push(item);
+        self.keys.insert(key, index);
+    }
+
+    pub fn get_id(&self, key: &str) -> usize {
+        self.keys[key]
+    }
+
+    pub fn get_item(&mut self, id: usize) -> &mut Item {
+        &mut self.items[id]
+    }
+}